feat(recommender): nightly ALS with offline metrics and CF candidates in the wave (M6)

- Implicit scores from play events and reactions per the signals table
  (90-day half-life, weaker skips outside the wave, dislikes override).
- numpy ALS for implicit feedback; time-split evaluation on the last
  7 days (NDCG@10, Recall@50) against a popularity baseline; a new model
  is activated only if NDCG@10 does not regress by more than 2%.
- Vectors are written to track_embeddings.cf and user_profiles_rec.taste_cf,
  versions to model_versions; NightlyTrainer runs at TRAIN_HOUR and on
  first start, `pipu-rec train` runs it manually.
- Wave: cf and item_cf candidate sources and a 0.10 cf term in the v1
  score; impressions are tagged heuristic-v1+cf.

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

services/recommender/pipu_rec/wave/service.py

@@ -9,7 +9,7 @@ import psycopg
 from pydantic import BaseModel, Field
 
 from pipu_rec.wave.contexts import Contexts, Rule
-from pipu_rec.wave.engine import MODEL_VER, Pick, build_batch, taste_centroids, update_session
+from pipu_rec.wave.engine import Pick, build_batch, model_version, taste_centroids, update_session
 from pipu_rec.wave.store import IndexHolder, SessionStore, load_profile, seed_tracks
 
 MAX_SERVED_MEMORY = 300
@@ -89,5 +89,6 @@ class WaveService:
         session.recent_artists = (session.recent_artists + artists)[-10:]
         self.sessions.put(req.session_id, session)
         return BatchResponse(
-            tracks=[BatchTrack(id=p.id, source=p.source, score=p.score) for p in picks], model_ver=MODEL_VER
+            tracks=[BatchTrack(id=p.id, source=p.source, score=p.score) for p in picks],
+            model_ver=model_version(index, profile),
         )