feat(embeddings): cache persistent de vectori in SQLite + warmup in fundal

Vectorii corpusului k-NN persista in tabela embedding_cache (PK model+text_hash,
blob float32 LE); la warmup se vectorizeaza doar textele lipsa din cache, deci
restartul cu corpus neschimbat nu mai plateste ~1-2 min de embed (embed=0).

- app/embedding_cache.py: serializare array('f'), load/save/purge chunk 500 cu
  BEGIN/COMMIT explicit (conexiuni autocommit), validare dimensiune la scriere
  si citire, orchestrare sync_corpus_vectors cu embed_fn injectat
- index_corpus(vectors=): vectori precalculati cu validare aliniere; mismatch
  -> fallback embed complet
- ensure_embeddings_corpus: warmup in thread la startup (block=True), calea de
  request ne-blocanta (acquire non-blocking pe lock; warmup in curs -> return
  imediat); purjare orfane + modele vechi doar dupa indexare reusita
- log warmup: cache=N embed=M in Xs
- 31 teste noi (cold/warm/incremental, model schimbat, concurenta, ranking
  exact, echivalenta float32); suita completa 1596 passed

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

app/main.py

@@ -7,6 +7,7 @@ un worker mort nu trebuie sa lase containerul "sanatos".
 from __future__ import annotations
 
 import secrets
+import threading
 from contextlib import asynccontextmanager
 from datetime import datetime, timezone
 from pathlib import Path
@@ -39,6 +40,25 @@ from .web.csrf import CsrfError
 from .web.session import AdminRequired, LoginRequired
 
 
+def _warmup_embeddings() -> None:
+    """Incarca modelul de embeddings si indexeaza corpusul SILVER, in fundal.
+
+    Ruleaza intr-un thread daemon la startup: incarcarea modelului (~230MB) plus
+    vectorizarea corpusului dureaza zeci de secunde si NU are voie sa blocheze
+    primul request pe /mapari. Pana termina, sugestiile embeddings lipsesc
+    (degradare gratioasa); GOLD/SILVER/fuzzy functioneaza normal.
+    """
+    from .mapping import ensure_embeddings_corpus
+    try:
+        conn = get_connection()
+        try:
+            ensure_embeddings_corpus(conn, block=True)
+        finally:
+            conn.close()
+    except Exception:
+        pass  # best-effort: esecul warmup-ului nu opreste API-ul
+
+
 @asynccontextmanager
 async def lifespan(app: FastAPI):
     install_log_redaction()
@@ -49,6 +69,8 @@ async def lifespan(app: FastAPI):
     # cheia API si secretul de sesiune, in loc de o instanta descoperita post-deploy.
     validate_prod_invariants(get_settings())
     init_db()
+    if get_settings().embeddings_enabled:
+        threading.Thread(target=_warmup_embeddings, name="emb-warmup", daemon=True).start()
     yield