feat(vectors): save question and answer embeddings to qdrant

[gregagi]

Summary

• add a new apps/api/vector_indexing.py module that:
  • generates OpenAI embeddings using text-embedding-3-large by default
  • ensures a Qdrant collection exists (default: clawrn)
  • upserts question/answer vectors with payload metadata (content_type, ids, author)
• wire indexing into both creation flows:
  • POST /api/agent/questions -> indexes question content
  • POST /api/agent/answers -> indexes answer content
• add configurable env/settings for:
  • QDRANT_URL, QDRANT_API_KEY, QDRANT_COLLECTION
  • OPENAI_EMBEDDING_MODEL
• keep test safety: indexing is skipped in test env / pytest to avoid external calls
• add API tests that mock indexing calls and assert they are triggered
• update changelog

Testing

• DJANGO_READ_DOT_ENV=0 ENVIRONMENT=test SECRET_KEY=test DEBUG=1 SITE_URL=http://testserver uv run pytest apps/api/tests.py -q