AI providers: review follow-ups to 0060ccd

Replay usage key gemini_usage -> llm_usage (backend + ReplayTab), re-price the
correction/correlation estimates for gpt-4o-mini, no temperature for o-series
or gpt-5 models, and refresh comments still naming Gemini.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

drb-c2-core/app/config.py

@@ -122,17 +122,17 @@ class Settings(BaseSettings):
     # NWS alerts cache TTL; the issue asks 60-120 s (floored at 30 s in code).
     weather_alerts_ttl_s: int = 90
 
-    # Gemini (intelligence extraction, embeddings, incident summaries)
+    # Gemini: only used if CORR_CHEAP_MODEL / TRANSCRIPT_CORRECTION_MODEL is set
+    # to a gemini-* ID (app/internal/llm.py). OpenAI serves both by default.
     gemini_api_key: Optional[str] = None
     # Correlation LLM (server-26#299). The rules engine decides every confident
     # case; this model is called at most ONCE per scene, and only when the rules
     # flag the decision as ambiguous (incident_correlator._ambiguity). There is
     # no second "smart" tier any more: the cheap/smart consensus tiebreak ran on
     # ~15% of scenes and resolved nearly every disagreement as "new" (#115).
-    # The model ID was once retired by Google and 404'd on every call for days;
-    # a failed LLM call falls back to the rules decision, so nothing broke
-    # loudly. Verify against https://ai.google.dev/gemini-api/docs/models
-    # before changing.
+    # A model ID was once retired by its provider and 404'd on every call for
+    # days; a failed LLM call falls back to the rules decision, so nothing broke
+    # loudly. Verify the ID against the provider's model list before changing.
     corr_cheap_model: str = "gpt-4o-mini"   # OpenAI since 2026-11-08 (was gemini-3.6-flash); a gemini-* ID still works
     # Transcript correction (server-26#36). Runs inside transcription, once per
     # transcribed call above MIN_WORDS_FOR_CORRECTION, so it is priced like STT