refactor(groups): take out the group's nostr identity and the curated lists, and keep broadcast unreachable

Lines B, D and H of docs/curated-to-mantra.md, pulled from the Curated fork this
morning, go again this afternoon: the group's nostr profile (kind 0), its four relay
lists (NIP-65, NIP-17, NIP-50, NIP-51), its posts (kind 1), the curated schemas it
publishes (31889), the suggestions it reads (31888) and the entries it accepts
(31890). Eighty-two files, twelve screens, the `nostr/curated/` package, the readings
(`GroupNostrProfile`, `GroupRelayList`, `GroupPost`, `GroupCuratedSchema`,
`GroupCuratedEntry`, `CuratedSuggestion`), the `applyInnerEvent` arms for all eight
kinds, the `ChatRepository` and `NostrRepository` reads that fed them, 226 strings,
and every test that came with them. A translation collective has no list to curate
and no reason to describe itself under a key no member controls, and five rows
saying so on every group's screen were five rows about somebody else's product.

**The pasted-event proposal goes with them, because it was theirs.**
`GroupEventProposal.ACCEPTED_KINDS` was exactly kinds 0, 1, the four relay-list
kinds and 31889 -- "the kinds the group's screen has a place for", in its own words --
so with the five rows gone it would have refused every paste. Keeping it as a
generic "sign any event" was considered and rejected: the screen's whole argument
was that a member goes to the row to see the event landed, and there is no row. The
`ProposedEvent` summaries for the same kinds go too; the one test of its pre-existing
fallback ("Event of kind N") is kept, as the only line of `ProposedEventTest` that
was about code this repository had before the pull.

**Broadcast stays, and nothing opens it.** `BroadcastGroupSignedEventScreen`, its
route, view model, state and `BroadcastButton` are kept, on the decision that a way
to send a group-signed event to relays is worth having against the day something
wires a button in -- the button's only call sites were the five removed screens.
Two things had to change for it to compile against a tree with no relay lists:
`defaultRelaysFor(kind)` is now the app's own publish set for every kind, since the
General list it preferred, the Blocked list it filtered by and the schema relays it
widened to no longer exist; and `relayUrlOrNull` moves into the view model's
companion from the deleted `GroupRelaySet`, unchanged. The hint on the screen says
"the relays this app publishes to" rather than "where the group has said it lives".
Its two tests are rewritten around the new seed: the screen test answers for the
first two seeded relays and counts the rest as asked, and empties the list by hand
to see the empty state, since the seed always has something in it. Deleting
broadcast outright -- the tidier tree -- was the recommendation and was declined.

**Every pre-existing file is back at its pre-pull content plus the kept lines'
hunks, and nothing else.** Eleven files -- `ChatMessage.kt`, `NostrEventDao.kt`,
`NostrEvent.kt`, `LocalChatRoom.kt`, `Member.kt`, `ProposedEvent.kt`,
`ChatTranscript.kt`, `ProfileAvatar.kt`, the group screen's view model and state,
and `m3-title-case.py` -- were touched by no kept commit and are restored from
ba0830a3 byte for byte, so `inComparableGroups` is a private helper of the group
screen again rather than a shared extension one deleted screen needed, and
`ProfileAvatar` has one overload again. The rest were restored and had the kept
hunks re-applied: the back button's three on `ChatRoomDetailScreen`, broadcast's
`groupSignedEvent` read on the two chat repositories, and on the two nostr
repositories the sign-in reads, by reverse-applying the queue commit's hunk. The
check is `git diff ba0830a3 -- <file>`, which shows only those. Reverting the eight
commits was rejected because the back button and the read-only identity work landed
on top of them and would have conflicted in every one of the shared files; editing
the current files by hand was rejected because it leaves residue that a diff against
the base cannot distinguish from a decision.

**The strings that went are exactly the ones nothing references any more and that
the pull added.** Six strings were unreferenced before the pull and stay; the header
sentence and one capitalised "Mantra" that the queue commit's join carried are prose,
not feature, and stay too. The seven section comments that described removed blocks
go; the broadcast block's stays.

**The plan's third decision said "hide the two rows behind a constant".** That
covered the curated rows and not the profile, relays and posts beside them, and the
call was to take the whole identity block out. The record of what went and what
stayed is the paragraph after the built table in docs/curated-to-mantra.md, with the
seam it leaves for the next pull: an upstream commit that touches the identity block
conflicts at `ChatRoomDetailScreen`, `MantraNavHost` and `strings.xml`, and is
dropped. docs/README.md says the same in a sentence. The nsec and npub notes still
name `EditGroupCuratedSchemaViewModel`; they are records of what was built upstream
and are left as written.

Verified with :composeApp:compileDebugKotlinAndroid, :composeApp:compileKotlinJvm,
:composeApp:jvmTest (826 tests, from 1,039), :composeApp:testDebugUnitTest (413,
from 530) and :composeApp:m3Audit, every budget met, 12 adaptive uses at the floor.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

docs/README.md

@@ -40,8 +40,10 @@ only one about what the app looks like rather than what it does; read the
 jvm-target note first if you want to know why its adaptive-layout phase exists.
 The curated-to-mantra note is a phased plan that has been built, save for its last
 phase, which is a decision: it is about the repository rather than the app, and reads
-alone, except that its first decision leans on the derivation note's one rule. The
-nsec and npub sign-in notes arrived with it and record what they built.
+alone, except that its first decision leans on the derivation note's one rule. Two of
+the lines it pulled -- the group's nostr identity and the curated lists -- were taken
+out again the same day, and its record says what went and what stayed. The nsec and
+npub sign-in notes arrived with it and record what they built.
 The nsec sign-in note is a phased plan that has been built; it inherits the
 key-storage decision from the jvm-target note and drives the navigation state
 machine `NavigationViewModel.processLocalAccount` implements, so read it with the

docs/curated-to-mantra.md

@@ -32,6 +32,31 @@ says so, and it did so in seven places worth reading:
 | the exactness diff is the logo plus `808a3459`'s three files | plus the fourteen files Mantra changed on its own side: Phase 0's prose at the protected names, `39fb64b6`'s four, and this document with its two scripts. Seventeen files, every one accounted for, at every cut |
 | 1,036 jvm tests at the end | 1,039 — the three from `39fb64b6` — and per phase 736, 864, 905, 1,007, 1,039; `testDebugUnitTest` 403, 486, 495, 530, 530; every phase clean on both compilers with every audit budget met |
 
+**Lines B and D were taken out again the same day**, and H with them, by a product
+decision rather than a merge problem: the group's nostr profile, its relay lists, its
+posts, the curated schemas and the curated entries are not Mantra's product, and the
+[third decision](#3-curated-lists-are-not-mantras-product-and-mantra-should-take-them-anyway)'s
+"hide the two rows behind a constant" was not enough once the profile, relays and
+posts rows went with them. What went: the five rows and the twelve screens behind
+them, `nostr/curated/`, the readings (`GroupNostrProfile`, `GroupRelayList`,
+`GroupPost`, `GroupCuratedSchema`, `GroupCuratedEntry`, `CuratedSuggestion`), the
+`applyInnerEvent` arms for kinds 0, 1, the four relay-list kinds, 31889 and 31890,
+the `ChatRepository` and `NostrRepository` reads that fed them, the pasted-event
+proposal (`GroupEventProposal`, whose accepted kinds were exactly those), the
+`ProposedEvent` summaries for the same kinds, 226 strings, and every test that came
+with them. What stayed from the pull: F (the back button), G and I (sign-in), J (the
+library chore), and E (broadcast) -- which nothing opens now, since its button sat on
+the five removed screens and its relay seeding read the group's relay lists and
+schemas; `defaultRelaysFor` is the app's publish set, and the route and the button
+are kept against the day something wires one in. Every pre-existing file the removed
+commits touched is back at its pre-pull content plus the kept lines' hunks; the
+exactness check for that is `git diff ba0830a3 -- <file>` showing only the back
+button, broadcast's `groupSignedEvent`, and the sign-in reads. Tests afterwards: 826
+jvm, 413 common, every audit budget met. The seam this leaves for the next pull is the
+one the third decision predicted: any upstream commit that touches the identity block
+will conflict at `ChatRoomDetailScreen`, `MantraNavHost` and `strings.xml`, and the
+resolution is to drop it.
+
 The upstream moved before this landed: `curated/curated` went from `86cb876b` to
 `29027f2b`, ten commits for several profiles on one device, one of them the fork's
 first schema migration (version 20). They are the next pull, and the schema change is

docs/scripts/m3-title-case.py

@@ -29,8 +29,6 @@ SMALL = {'a', 'an', 'the', 'to', 'of', 'for', 'and', 'or', 'via',
 SAMPLE = {
     'Steve Biko', 'John Doe', 'Frank Talk', 'Alan Turing',
     'To Kill a Mockingbird', 'Man With A Plan', 'Woman Of Few Words',
-    # Films, as suggested to a curated list in the queue screen's preview.
-    'The Rise and Rise of Bitcoin', 'Magic Money',
 }