dsh-model-reasoning-effort: Restoring reasoning effort editing for custom pi-ai models

The official DSH "Models" settings page does not provide editing for `reasoningEfforts`, although the pi-ai adapter itself supports this field. The `dsh-model-reasoning-effort` plugin (MIT, v0.1.0) maintained by pwkazn adds a "Reasoning Effort (Custom Models)" panel at the bottom of the Settings → Models page. This is implemented by registering the `settings.models.footer` slot and does not interfere with the `provider-card` key. It supports a three-state toggle per model (Inherit / No Reasoning / Declared Effort), separating the effort level from the value sent, and allows for batch writing of all models belonging to the same provider. It adheres to adapter validation rules before writing. The write operation uses `settings.mutate` to replace the entire `models` array and includes a `revision` to prevent concurrency issues. The declaration is an assertion rather than a probe; effort levels not supported by the gateway will fail at runtime. The panel only edits the user-level `llm-pi-ai.providers.<route>.models` and does not affect the DeepSeek direct adapter. Installation command: `dsh plugin --profile web add git+https://github.com/pwkazn/dsh-model-reasoning-effort.git`