fix: start the app recommendations flow eagerly to pin notes in the soft cache

With WhileSubscribed, nothing held a strong reference to my own 31989
notes while no UI was collecting; under memory pressure the soft cache
could drop them, and the next recommendApp would rebuild that kind's
event from an empty snapshot, wiping previously recommended apps.
Eagerly keeps the observer registered (its result set strongly
references the notes), matching the sibling account states.

https://claude.ai/code/session_015dX5vWqvXUYD8rzPYX8vTB

amethyst/src/main/java/com/vitorpamplona/amethyst/model/nip89AppHandlers/AppRecommendationsState.kt

@@ -64,13 +64,19 @@ class AppRecommendationsState(
      * My kind 31989 recommendation events (one per handled kind), kept in
      * sync as the cache consumes new versions. UI should collect this
      * instead of rescanning the cache on every event bundle.
+     *
+     * Eagerly started on purpose, like the sibling account states: the
+     * registered observer holds strong references to these notes, pinning
+     * them in the soft-reference cache so the read-modify-write publishers
+     * below never rebuild a 31989 from a partially garbage-collected
+     * snapshot (which would silently drop previously recommended apps).
      */
     val flow: StateFlow<List<AppRecommendationEvent>> =
         cache
             .observeEvents<AppRecommendationEvent>(
                 Filter(kinds = listOf(AppRecommendationEvent.KIND), authors = listOf(signer.pubKey)),
             ).flowOn(Dispatchers.IO)
-            .stateIn(scope, SharingStarted.WhileSubscribed(30000), existingRecommendationEvents())
+            .stateIn(scope, SharingStarted.Eagerly, existingRecommendationEvents())
 
     /**
      * Serializes read-modify-write of the per-kind recommendation events so