Switch from fastembed to Python sentence-transformers for embedding

ort (ONNX Runtime) has no prebuilt binaries for aarch64-musl.
Use a Python subprocess with sentence-transformers instead:
- scripts/embed.py: reads JSON stdin, outputs embeddings
- kb.rs: calls Python script via tokio subprocess
- Dockerfile: install python3 + sentence-transformers

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

scripts/embed.py

@@ -0,0 +1,26 @@
+#!/usr/bin/env python3
+"""Generate embeddings for text chunks. Reads JSON from stdin, writes JSON to stdout.
+
+Input:  {"texts": ["text1", "text2", ...]}
+Output: {"embeddings": [[0.1, 0.2, ...], [0.3, 0.4, ...], ...]}
+"""
+import json
+import sys
+from sentence_transformers import SentenceTransformer
+
+MODEL_NAME = "all-MiniLM-L6-v2"
+
+def main():
+    data = json.loads(sys.stdin.read())
+    texts = data["texts"]
+
+    if not texts:
+        print(json.dumps({"embeddings": []}))
+        return
+
+    model = SentenceTransformer(MODEL_NAME)
+    embeddings = model.encode(texts, normalize_embeddings=True)
+    print(json.dumps({"embeddings": embeddings.tolist()}))
+
+if __name__ == "__main__":
+    main()