Add post-diarization speaker review workflow

PROJECT_KNOWLEDGE.md

@@ -81,7 +81,8 @@ Imported audio
        (optionally normalize loudness; default on)
     -> transcribe with whisper.cpp and large-v3-turbo
     -> optionally diarize with pyannote.audio Community-1
-    -> generate a protocol directly from the full transcript
+    -> when diarized, wait for optional speaker review before explicit protocol generation
+    -> otherwise generate a protocol directly from the full transcript
     -> review and edit by a human
 ```
 
@@ -121,7 +122,10 @@ Speaker selectors filter participants using a snapshot of all current widget
 selections, falling back to saved mappings before first interaction. Filtering
 retains existing assignments; clearing a selection releases the participant.
 Completeness counts and warnings cover detected speakers only and do not gate
-protocol regeneration or change mapping persistence.
+protocol generation or change mapping persistence. A diarized run persists the
+`awaiting_speaker_review` checkpoint after diarization; an empty, partial or
+complete mapping can then explicitly generate a protocol from the existing
+artifacts. Failed generation preserves that checkpoint and saved mappings for retry.
 
 ## Progress Contract