REDI-101: remove AI/tooling attribution from docs & source comments

Make the repo read as a hand-authored reference project by dropping
references to the authoring tooling — the internal convention "skills", the
Linear backlog, and the REDI issue ids — from the README and source comments.

- README: remove the "Convention skills index" section and its TOC entry,
  strip every `See **android-...**` / koin-constructor-dsl citation, drop the
  Linear backlog link, and reword the REDI-99 Room reference to "an optional
  stretch".
- Source comments: neutralize the four skill citations in HttpClientExt,
  CharactersPresentationModule, CharacterListRobot, ErrorDemoViewModel.
- Also correct the testing section that the MockK migration left stale: it
  described "Fakes, not mocks" / FakeCharacterRepository (now deleted) — now
  describes MockK, and adds MockK to the stack lists.

Verified: git grep finds no AI/Claude/Anthropic/Linear/skill references in any
tracked file; assembleDebug + assembleDebugAndroidTest green.

feature/characters/presentation-compose/src/androidTest/kotlin/com/example/architecture/feature/characters/presentation/compose/CharacterListRobot.kt

@@ -14,7 +14,7 @@ import org.junit.Assert.assertTrue
  * Robot for [CharacterListScreen] UI tests. Each method returns `this` so calls read as a fluent
  * scenario (`robot.setContent(state).assertCharacterShown(...).clickCharacter(...)`). The robot owns
  * the interaction vocabulary; the test owns the assertions' intent — keeping tests readable and
- * resilient to UI structure changes. See android-testing.
+ * resilient to UI structure changes.
  */
 class CharacterListRobot(
     private val composeRule: ComposeContentTestRule,