Try using more open embedding models

[vitawasalreadytaken]

At the moment we're using OpenAI's text-embedding-3-large model for embedding document contents. It would be nice to migrate to more open models that can also be self-hosted. Some candidates could be:

• jina-embeddings-v4 or v3; we've had some success with v2 in the past but then OpenAI proved to be better.
• Apertus
• lots of other choices from the MTEB Leaderboard – perhaps this is the best place to start.