correlator: LLM tier reads the scene's transcript, not the whole call (server-26#102)

The last leg of the #80/#95 scene-context leak. llm_correlator._call_block
read call_doc's whole-call transcript for every scene, so on a multi-scene
call every scene's cheap-tier and tiebreaker decision was made against text
that also contained the other scenes.

- intelligence.py: each processed[] scene now carries its own "transcript" —
  transcript_corrected, else this scene's segments joined, else (single scene)
  the whole transcript.
- _build_context / preview_correlation / correlate_call: take a `transcript`
  param; _build_context resolves ctx["scene_transcript"] from it, falling
  back to the call doc (sweep, single-scene, tests) — the fallback is kept
  here, unlike embedding/severity, because a scene always has real text.
- upload.py: both scene loops pass scene["transcript"].
- llm_correlator._call_block: reads ctx["scene_transcript"] (call-doc
  fallback retained for test-built ctx).
- recorrelation_sweep: passes the call doc's text explicitly.
- +1 regression test. Full c2-core suite green (296 passed, sandboxed venv).

NOT for merge until the running correlation measurement window closes and its
dump is analysed — deploying a correlator change mid-window would mix old and
new behaviour in the sample.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

drb-c2-core/app/internal/intelligence.py

@@ -172,7 +172,7 @@ async def extract_scenes(
 
     Each scene dict contains:
         tags, incident_type, location, location_coords, resolved,
-        severity, vehicles, units, transcript_corrected,
+        severity, vehicles, units, transcript, transcript_corrected,
         segment_indices, embedding
 
     Side-effect: updates calls/{call_id} in Firestore with merged tags,
@@ -337,6 +337,19 @@ async def extract_scenes(
         )
         embedding = await asyncio.to_thread(_sync_embed, scene_text)
 
+        # This scene's own words, unprefixed — corrected text if we have it,
+        # else the raw segments this scene owns, else (single-scene) the whole
+        # transcript. The correlator's LLM tier reads this per scene instead of
+        # the call doc's whole-call transcript (server-26#102).
+        if transcript_corrected:
+            scene_transcript = transcript_corrected
+        elif segments and segment_indices:
+            scene_transcript = " ".join(
+                segments[i]["text"] for i in segment_indices if i < len(segments)
+            )
+        else:
+            scene_transcript = transcript
+
         processed.append({
             "tags":                 tags,
             "incident_type":        incident_type,
@@ -348,6 +361,7 @@ async def extract_scenes(
             "severity":             severity,
             "resolved":             resolved,
             "reassignment":         reassignment,
+            "transcript":           scene_transcript,
             "transcript_corrected": transcript_corrected,
             "segment_indices":      segment_indices,
             "embedding":            embedding,