LocalMind 1.0.22

Centralize local model discovery and status polling

Introduce a shared `ModelCatalog` service that periodically refreshes all local providers, tracks available models, and publishes per-provider availability/model snapshots. Update `ILocalModelProvider` to return `LocalProviderStatus` via `GetStatusAsync`, and adapt Foundry, Ollama, and llama.cpp providers to the new contract.

Refactor `MainViewModel`, `ChatViewModel`, and `SettingsViewModel` to consume catalog-driven updates instead of doing their own provider polling. This keeps chat model lists and settings provider states in sync as providers come online/offline, while preserving selection stability and reducing duplicated refresh logic.