Add AI stack configuration: ELAI avatars, system params, training tips import

- Add ELAI avatar seed data (7 avatars with codes, URLs, voice configs)
  from the original backend's avatars.json
- Add missing system parameters: encoach.aws_region (eu-west-1),
  encoach.whisper_workers (4)
- Add training tips import script with pathways_2_rw.json data source
- Add action_compute_embeddings() method to training tip model for
  computing sentence-transformer embeddings on demand

Made-with: Cursor

scripts/import_training_tips.py

@@ -0,0 +1,76 @@
+"""Import training tips from pathways JSON into encoach.training.tip model.
+
+Run inside the Odoo container via:
+    odoo shell -d encoach < /mnt/custom/scripts/import_training_tips.py
+
+This script:
+  1. Reads pathways_2_rw.json
+  2. Creates encoach.training.tip records (tip_id, category, content)
+  3. Computes sentence-transformer embeddings and stores them
+"""
+import json
+import os
+import pickle
+import sys
+
+SCRIPT_DIR = os.path.dirname(os.path.abspath(__file__))
+JSON_PATH = os.path.join(SCRIPT_DIR, "pathways_2_rw.json")
+
+
+def main():
+    with open(JSON_PATH, "r", encoding="utf-8") as f:
+        data = json.load(f)
+
+    TipModel = env["encoach.training.tip"].sudo()
+
+    tips_data = []
+    for unit in data["units"]:
+        for page in unit["pages"]:
+            for tip in page["tips"]:
+                category = tip["category"].lower().replace(" ", "_")
+                tips_data.append({
+                    "tip_id": tip["id"],
+                    "category": category,
+                    "content": tip["text"],
+                })
+
+    created = 0
+    skipped = 0
+    for td in tips_data:
+        existing = TipModel.search([("tip_id", "=", td["tip_id"])], limit=1)
+        if existing:
+            skipped += 1
+            continue
+        TipModel.create(td)
+        created += 1
+
+    env.cr.commit()
+    print(f"Tips import complete: {created} created, {skipped} skipped (already exist)")
+
+    try:
+        from sentence_transformers import SentenceTransformer
+        import numpy as np
+
+        print("Computing embeddings with all-MiniLM-L6-v2...")
+        model = SentenceTransformer("all-MiniLM-L6-v2")
+
+        tips = TipModel.search([("embedding", "=", False)])
+        if not tips:
+            print("All tips already have embeddings.")
+            return
+
+        for i, tip in enumerate(tips):
+            vec = model.encode([tip.content]).astype(np.float32)[0]
+            tip.embedding = pickle.dumps(vec)
+            if (i + 1) % 10 == 0:
+                print(f"  Embedded {i + 1}/{len(tips)} tips...")
+
+        env.cr.commit()
+        print(f"Embeddings computed for {len(tips)} tips.")
+
+    except ImportError:
+        print("WARNING: sentence-transformers not available. Tips created without embeddings.")
+        print("Embeddings can be computed later by re-running this script.")
+
+
+main()