fix: classifier robustness — schema in prompt, retries, kickoff fallback

The kickoff parse-and-confirm route was 500-ing intermittently because
Hermes-3 + Featherless's response_format={"type":"json_object"} only
guarantees JSON output, NOT a particular schema. The model was inventing
its own field names (sceneTime, entities, settingDetails) instead of
the KickoffParse fields, causing Pydantic validation to fail on both
classify() retries.

Three changes:

1. Include the Pydantic JSON schema in the system prompt so the model
   knows exactly which keys to produce. Affects every classify() call
   (kickoff parse, turn parse, scene-close detect, significance,
   state-update, scene summarize). Strip ```json fences if the model
   wraps its output. Bump retries 2 → 3 (model is stochastic; one extra
   attempt closes most of the remaining gap).

2. parse_kickoff() now passes a default empty KickoffParse so the
   route degrades to a fillable form instead of 500 when the classifier
   ultimately fails. The confirm form is the human-in-the-loop; an
   empty form is strictly better UX than a stack trace.

3. Tests updated: bumped canned-failure arrays from 2 → 3 entries to
   match the new attempt count; renamed kickoff test from
   "raises_when_classifier_fails_twice" to
   "falls_back_to_empty_when_classifier_fails" reflecting the new
   degraded-but-usable behavior.

Verified live with all 3 sample bots (maya/eli/sam) — kickoff route
returns 200 across multiple attempts. Full suite: 168 passed.

tests/test_classify.py

@@ -18,7 +18,7 @@ async def test_classify_parses_valid_json():
 
 @pytest.mark.asyncio
 async def test_classify_falls_back_on_unparseable_after_retry():
-    mock = MockLLMClient(canned=["nope", "still nope"])
+    mock = MockLLMClient(canned=["nope", "still nope", "nope3"])
     default = Verdict(score=1, reason="fallback")
     result = await classify(mock, model="m", system="x", user="y", schema=Verdict, default=default)
     assert result.reason == "fallback"