AI providers: correlation decision and transcript correction move to OpenAI gpt-4o-mini

One funded account (OpenAI) now covers extraction, embeddings, summaries,
the ambiguous-scene correlation call and transcript correction; Groq stays
optional for STT. New app/internal/llm.py picks the provider from the model
name, so setting either model back to a gemini-* ID needs no code change.
Usage accounting for replay runs now covers both providers.

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

drb-c2-core/tests/test_ai_timeouts.py

@@ -164,7 +164,7 @@ async def test_gemini_timeout_in_decide_falls_back_to_rules():
     def hang(model, prompt):
         raise httpx.ReadTimeout("The read operation timed out")
 
-    with patch.object(settings, "gemini_api_key", "test-key"), \
+    with patch.object(settings, "openai_api_key", "test-key"), \
          patch.object(llm_correlator, "_build_decide_prompt", lambda ctx, cands: "p"), \
          patch.object(llm_correlator, "_sync_gemini", hang):
         assert await llm_correlator.decide("c1", _corr_ctx()) is None  # None == keep the rules decision
@@ -179,7 +179,7 @@ async def test_gemini_timeout_in_transcript_correction_keeps_raw_text():
         raise httpx.ReadTimeout("The read operation timed out")
 
     text = "engine 5 respond to a structure fire at 12 main street smoke showing"
-    with patch.object(settings, "gemini_api_key", "test-key"), \
+    with patch.object(settings, "openai_api_key", "test-key"), \
          patch.object(transcript_correction, "resolve_context", AsyncMock(return_value={})), \
          patch.object(transcript_correction, "_sync_gemini", hang):
         assert await transcript_correction.correct("c1", text) == (None, None, False)