docs: hosted openai-compatible client example (name:model id)

Is your feature request related to a problem? Please describe.

config.example.yaml shows type: openai-compatible for Ollama/localhost. A hosted Chat Completions host uses the same client type, with two easy misses: model must be name:catalog-id, and api_base is the /v1 root (not /v1/chat/completions). Embedding rows in the example do not apply to a chat-only host.

Describe the solution you'd like

Working ~/.config/aichat/config.yaml for a live OpenAI-compatible host (PZERO). Public catalog: GET https://api.pzero.studio/v1/models. Default text id: deepseek-v4-flash. Prepaid Bearer key (pzero_…).

model: pzero:deepseek-v4-flash
clients:
  - type: openai-compatible
    name: pzero
    api_base: https://api.pzero.studio/v1
    api_key: pzero_…
    models:
      - name: deepseek-v4-flash

aichat --model pzero:deepseek-v4-flash "ping"

Do not add type: embedding models against this host. name must not contain dots or colons. model prefix must match name.

No first-party bundled provider asked.

Describe alternatives you've considered

Using type: openai with only api_base overridden. The example file already documents openai-compatible for non-OpenAI hosts, so that is the right seam.

Additional context

Wiki: https://github.com/sigoden/aichat/wiki/Configuration-Guide Example: https://github.com/sigoden/aichat/blob/main/config.example.yaml