docs: OPENAI_BASE_URL must include /v1 because the app concatenates /models

Describe your feature request

Chat UI already speaks only OpenAI-compatible APIs via OPENAI_BASE_URL and discovers models from ${OPENAI_BASE_URL}/models (local install). Two details are easy to get wrong on a hosted Chat Completions host:

1. OPENAI_BASE_URL must include /v1. The app concatenates /models. Origin-only 404s. Do not append /chat/completions.
2. Do not point embeddings or web-search helpers at a chat-only host. Chat UI already dropped those provider-specific extras.

PZERO is a live OpenAI-compatible host. Catalog: GET https://api.pzero.studio/v1/models (public). Default text id: deepseek-v4-flash. Prepaid Bearer key (pzero_…).

OPENAI_BASE_URL=https://api.pzero.studio/v1
OPENAI_API_KEY=pzero_…

Distinct from Hugging Face Hub docs. This is the chat-ui OPENAI_BASE_URL concat gotcha.

Implementation idea

No first-party provider asked. If a named row in the provider table is too much, the /v1 concat gotcha is the docs gap.