fix(together-ai): use a valid Together embedding model as default

[ankit25bcs10610]

What

The Together AI embed config defaults the model to mistral-embed:

model: {
  param: 'model',
  required: true,
  default: 'mistral-embed',
},

This is copy-pasted from mistral-ai/embed.ts. Together AI does not serve mistral-embed, so an embeddings request that relies on the default (omits model) is sent an invalid model id and fails.

Fix

Default to BAAI/bge-large-en-v1.5 — the embedding model used in Together's own Create Embedding API docs example.

Note: Together's hosted model list shifts over time. I picked the model from their official docs example; maintainers may prefer a different supported default (e.g. togethercomputer/m2-bert-80M-8k-retrieval) — happy to change it.

npm run format:check passes.