[FEAT] User Onboarding Improvements (Configurable User parameters)

[NeonDaniel]

Objective

Currently, user profile fields are required during account creation. A potentially better alternative would be to move this to a separate modal and optional UI flow.

Initial Implementation Requirements

• Define a configuration structure for specifying profile fields (possibly with configurable validation)
• Update profile endpoints to accept inputs for available fields
• Design an LLM-powered onboarding flow
  • This may be implemented using MCP tools to parse configured fields to determine what to ask the user
  • Responses may be parsed using guided decoding to ensure a valid config option is extracted from user messages (or trigger clarification if no valid option was extracted)

Other Considerations

• Example implementation in this commit
• Config fields may be configured with a JSON Schema or similar to specify valid options for each field
• The LLM-powered onboarding may be split to a separate issue
• Multiple elements to this; the user creation UI flow also needs to be updated to support different applications

[NeonDaniel]

• Config fields may be configured with a JSON Schema or similar to specify valid options for each field

Example for how to build a Pydantic object from JSON schema https://github.com/NeonGeckoCom/brainforge-context-server/blob/40083284d5aec85204413f806f5910316ce76e20/brainforge_context_server/util.py#L5-L23

[NeonDaniel]

Related to/dependent on #26