feat: módulo embedding — encode, model CRUD, stores sqlvec y usearch

Funciones Python para embeddings: carga/guardado de modelos, encoding de
texto, y almacenamiento/búsqueda vectorial con sqlite-vec y usearch.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

python/functions/embedding/embedding_search_sqlvec.md

@@ -0,0 +1,37 @@
+---
+name: embedding_search_sqlvec
+kind: function
+lang: py
+domain: infra
+version: "1.0.0"
+purity: impure
+signature: "def embedding_search_sqlvec(db_path: str, table: str, query_embedding: list, k: int = 10) -> list"
+description: "Busca los k vecinos mas cercanos en tabla sqlite-vec. Retorna rowids y distancias ordenados."
+tags: [embedding, sqlite, vector, search, retrieval, sqlite-vec, python]
+uses_functions: []
+uses_types: []
+returns: []
+returns_optional: false
+error_type: "error_go_core"
+imports: [sqlite3, sqlite_vec, numpy]
+tested: false
+tests: []
+test_file_path: ""
+file_path: "python/functions/embedding/sqlvec.py"
+---
+
+## Ejemplo
+
+```python
+model = embedding_load_model(".local/models/e5-small")
+q_emb = embedding_encode(model, ["¿Que es machine learning?"], mode="query")[0]
+
+results = embedding_search_sqlvec("vectors.db", "doc_embeddings", q_emb, k=5)
+# [{"rowid": 0, "distance": 0.23}, {"rowid": 1, "distance": 0.45}, ...]
+```
+
+## Notas
+
+Busqueda brute-force (exacta, no aproximada). Para 50k vectores tarda ~19ms/query.
+El campo distance es distancia coseno (menor = mas similar) porque los embeddings estan normalizados.
+Cold start rapido (~18ms) porque SQLite no carga todo el indice a RAM.