docs(llm): clarify the function body-format contract for OpenAI-compatible routes

Problem

The self-managed LLM function workflow accepted CUSTOM for the function body
format while an otherwise equivalent request using OPENAI_CHAT was rejected.
The accepted API contract is not clear enough in the CLI examples and user
documentation, even though invocation uses the OpenAI-compatible
/v1/chat/completions route.

This ambiguity makes a valid router test look like an infrastructure failure
before the function is created.

Expected behavior

The API schema, CLI examples, and self-managed LLM examples agree on the
supported body-format value for OpenAI-compatible LLM functions. If both
values are intended, both work consistently; if only one is intended, the
unsupported value produces a precise validation error.

Acceptance criteria

• Confirm the intended body-format enum contract for LLM functions.
• Update CLI and self-managed examples to use the supported value.
• Explain the relationship between the function body format and the
  OpenAI-compatible invocation routes.
• Add request-validation coverage for accepted and rejected enum values.
• Ensure the rejection identifies the invalid field and accepted values.

By submitting this issue, you acknowledge that you are an assigned member of
the NVCF development team and agree to follow the project code of conduct and
contributing guidelines.

Start by locating the API schema, CLI examples, and self-managed LLM examples that define or use the function body-format value. Compare CUSTOM and OPENAI_CHAT with the request-validation entry points, then add coverage for accepted and rejected enum values. Done means the contract, examples, route relationship, and invalid-field error consistently agree.