feat: add custom model support for Vertex AI provider

[roomote]

Fixes #5751

Summary

This PR adds support for custom model input in the Vertex AI provider settings, allowing users to specify custom model IDs in the correct Vertex format (e.g., claude-sonnet-4@20250514) to avoid 404 errors.

Changes Made

UI Changes

• Vertex Provider Settings: Added a custom model input field with placeholder text showing the correct format
• Internationalization: Added translation keys for custom model UI elements

Backend Changes

• Provider Settings Schema: Extended vertexSchema to include optional vertexCustomModelId field
• Model Selection Logic: Updated useSelectedModel hook to prioritize custom models when provided
• AnthropicVertexHandler: Enhanced to handle custom model names with proper thinking suffix support

Testing

• Comprehensive Test Coverage: Added 25 tests for AnthropicVertexHandler custom model functionality
• Frontend Hook Tests: Added 8 tests for useSelectedModel hook with React Query mocking
• All tests passing: Both custom functionality and existing functionality verified

Technical Details

Model Format Handling

• Supports Vertex AI model format with @ separator (e.g., claude-sonnet-4@20250514)
• Automatically handles thinking models with :thinking suffix
• Maintains backward compatibility with predefined models

Validation

• Input field trims whitespace automatically
• Falls back to predefined models when custom model is empty
• Preserves existing model selection behavior for other providers

Testing Results

✅ AnthropicVertexHandler tests: 25/25 passing
✅ useSelectedModel hook tests: 8/8 passing
✅ All linting and type checking passed

User Experience

Users can now:

1. Navigate to Settings → Providers → Vertex AI
2. Enter a custom model ID in the "Custom Model" field
3. Use the correct Vertex format (e.g., claude-sonnet-4@20250514)
4. Avoid 404 errors when using newer or custom Vertex AI models

This resolves the core issue where users were getting 404 errors due to model name format mismatches between the UI suggestions and Vertex AI requirements.

---

Important

Adds custom model support for Vertex AI provider, allowing users to specify custom model IDs, with updates to UI, backend logic, and comprehensive testing.

• Behavior:
  • Adds custom model input support in Vertex AI provider settings, allowing custom model IDs like claude-sonnet-4@20250514.
  • Updates useSelectedModel hook to prioritize custom models.
  • Enhances AnthropicVertexHandler to handle custom model names with thinking suffix.
• Schema and Logic:
  • Extends vertexSchema in provider-settings.ts to include vertexCustomModelId.
  • Updates getSelectedModel in useSelectedModel.ts to handle custom models.
• UI:
  • Adds custom model input field in Vertex.tsx with placeholder and description.
  • Adds translation keys for custom model UI elements in settings.json.
• Testing:
  • Adds 25 tests for AnthropicVertexHandler custom model functionality in anthropic-vertex.spec.ts.
  • Adds 8 tests for useSelectedModel hook in useSelectedModel.spec.tsx.
  • All tests passing, ensuring both custom and existing functionality work correctly.

^{This description was created by for 56c872d. You can customize this summary. It will automatically update as commits are pushed.}

[ellipsis-dev]

The custom model handling here simply trims and returns the custom ID using fallback info, but it doesn’t strip a ':thinking' suffix as done in AnthropicVertexHandler.getModel. For consistency in UI display (and to avoid potential confusion), consider mirroring that logic.

Suggested change

	id: customModelId.trim(),
	id: customModelId.trim().replace(/:thinking$/, ""),

[ellipsis-dev]

The new custom model text field looks well-integrated with proper translation keys and placeholder text. If possible, consider externalizing the placeholder example ('claude-sonnet-4@20250514') into a translation key for consistency with internationalization practices.

Suggested change

	placeholder="claude-sonnet-4@20250514"
	placeholder={t("settings:placeholders.customModelExample")}

^{This comment was generated because it violated a code review rule: irule_C0ez7Rji6ANcGkkX.}