fix(worker): don't set cdn_hls_base until HLS is uploaded; add reindex

Two bugs fixed:

1. handle_transcode was writing cdn_hls_base = "hls/{version_id}" to the DB
   even though HLS files were only in a temp dir (never uploaded to Nextcloud).
   The stream endpoint then tried to serve this non-existent path, returning 404
   and breaking audio playback for every transcoded version. Removed the
   cdn_hls_base write — stream endpoint falls back to nc_file_path (raw file),
   which works correctly.

2. Added extract_peaks worker job type: lightweight job that downloads audio
   and computes waveform_peaks + waveform_peaks_mini only. No transcode, no HLS,
   no full analysis.

3. Added POST /internal/reindex-peaks endpoint (protected by internal secret):
   finds all audio_versions with null waveform_peaks and enqueues extract_peaks
   jobs. Safe to call multiple times. Use after a fresh DB scan or peak algorithm
   changes.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

api/src/rehearsalhub/routers/internal.py

@@ -10,11 +10,12 @@ from sqlalchemy.ext.asyncio import AsyncSession
 
 from rehearsalhub.config import get_settings
 from rehearsalhub.db.engine import get_session
-from rehearsalhub.db.models import BandMember, Member
+from rehearsalhub.db.models import AudioVersion, BandMember, Member
 from rehearsalhub.repositories.audio_version import AudioVersionRepository
 from rehearsalhub.repositories.band import BandRepository
 from rehearsalhub.repositories.rehearsal_session import RehearsalSessionRepository
 from rehearsalhub.repositories.song import SongRepository
+from rehearsalhub.queue.redis_queue import RedisJobQueue
 from rehearsalhub.schemas.audio_version import AudioVersionCreate
 from rehearsalhub.services.session import extract_session_folder, parse_rehearsal_date
 from rehearsalhub.services.song import SongService
@@ -148,3 +149,37 @@ async def nc_upload(
     )
     log.info("nc-upload: registered version %s for song '%s'", version.id, song.title)
     return {"status": "ok", "version_id": str(version.id), "song_id": str(song.id)}
+
+
+@router.post("/reindex-peaks", status_code=200)
+async def reindex_peaks(
+    session: AsyncSession = Depends(get_session),
+    _: None = Depends(_verify_internal_secret),
+):
+    """Enqueue extract_peaks jobs for every audio_version that has no waveform_peaks yet.
+
+    Safe to call multiple times — only versions with null peaks are targeted.
+    Useful after:
+    - Fresh DB creation + directory scan (peaks not yet computed)
+    - Peak algorithm changes (clear waveform_peaks, then call this)
+    - Worker was down during initial transcode
+    """
+    result = await session.execute(
+        select(AudioVersion).where(AudioVersion.waveform_peaks.is_(None))  # type: ignore[attr-defined]
+    )
+    versions = result.scalars().all()
+
+    if not versions:
+        return {"status": "ok", "queued": 0, "message": "All versions already have peaks"}
+
+    queue = RedisJobQueue(session)
+    queued = 0
+    for version in versions:
+        await queue.enqueue(
+            "extract_peaks",
+            {"version_id": str(version.id), "nc_file_path": version.nc_file_path},
+        )
+        queued += 1
+
+    log.info("reindex-peaks: queued %d extract_peaks jobs", queued)
+    return {"status": "ok", "queued": queued}