fix(provider): Add sync button and resolve bug merge models

[JamesDAdams]

Summary

• Add sync button on add/edit provider for check if a new models is available
• Fix duplicate models on Sync: In ProviderModal.tsx, fetchModels now claims both the merged model's id and each member's apiModelId. This prevents raw catalog models fetched from the provider backend from duplicating already configured and mode-merged models in the models list and in the Selected Models view.
• Preserve model selection during Sync: Newly discovered models fetched from the provider catalog are added to models and displayed in Available Models (unselected) without polluting Selected Models.
• Constrain mode suffix detection and parsing: Constrained splitModeSuffix to recognized MODE_SUFFIXES to prevent unrelated multi-hyphen model names from being grouped into pseudo-mode families.
• Fix Ollama tag preservation: Ensured parseModelValue validates against known reasoning effort values before splitting on a colon, preventing Ollama model tags (such as :latest, :instruct, :32b) from being stripped and treated as reasoning efforts.
• Preserve model metadata on update: Ensured updateModelSettings preserves existing model metadata (modes, requestBody, selected).

AI-Enhanced Development

Tell what models helped shape this PR:

• AI Models: Gemini 3.7 Flash

Cache Impact

Does this PR affect anything cached — system prompts, tool definitions, skills, or other context?

• No

[co-l]

Merged, will be released in v2.0.136

AI-summary

• Reviewed the sync/model-dedup changes and fixed 2 i18n lint errors that would have failed CI (Sync button label and its title now go through t() with en + fr)
• Verified the full unit suite (5087 passed), typecheck and lint clean, and the provider-rest/model-selection e2e green
• Rebased the PR onto latest develop and pushed the branch to the contributor's fork so it merges cleanly
• Squash-merged