AI provider adapter for embeddings

New Feature / Enhancement Checklist

• Report security issues confidentially.
• Any contribution is under this license.
• Before posting search existing issues.

Current Limitation

There is no standard way to configure an AI provider on the server. Everyone wires up their own SDK in Cloud Code, and server side features that need embeddings (like auto-embedding, #10606) have no provider to call.

Feature / Enhancement Description

An adapter in the existing pattern (like MailAdapter): configure a provider once in the server options, exposed in Cloud Code as something like Parse.AI.embed(). Embeddings are the priority since server internals need them; completions could be added later if there is demand, since you can already npm install any SDK in Cloud Code for that.

Example Use Case

1. Configure the adapter with a provider (OpenAI, Anthropic, local model, whatever implements the interface)
2. Call Parse.AI.embed(text) in Cloud Code
3. Auto-embedding (#10606) uses the same adapter internally

Alternatives / Workarounds

Install a provider SDK directly in Cloud Code. Fine for app code, but nothing inside Parse Server can rely on it.

3rd Party References

• Supabase AI / Supabase.ai.Session
• The Parse Dashboard AI panel has the same provider config need (parse-community/parse-dashboard#3256)

Blocks #10606

Start by examining the existing MailAdapter pattern and how server options are configured, then review the proposed Parse.AI.embed() entry point and the dependency on auto-embedding in #10606. Define what the provider interface and configuration need to support; done means a configured provider can serve embeddings to Cloud Code and the auto-embedding work.