feat: provider and model configuration system. and fix all linting errors

[ErikKoning (Erik-Koning)]

• Add ability to create custom providers and add multiple models per (retain big 3 model providers as default)
• Refactor default providers to use same same system as custom providers.
• Support multiple model configurations per provider with deployment-style selection
• Delay switching model until after task completes with UI indication
• Store custom user providers in ~/.openwork/user-providers.json
• Add ProviderConfigDialog for managing API keys, endpoints, and deployments
• Add AddProviderDialog for creating new custom providers
• Add explicit return types to all component functions
• Update UI components with proper TypeScript annotations
• Fix React hooks exhaustive-deps and setState-in-effect issues
• Fix 138 linting errors (missing return types, unused variables, etc.)

Description

Enables a unified system for managing and having users add their own providers or models locally. Default models and providers unchanged.
UI bug fixes, UI improvements
Lint errors solved

Related Issue

Fixes #(issue number)
#11 #7 #5

Type of Change

• Bug fix (non-breaking change that fixes an issue)
• New feature (non-breaking change that adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)
• Documentation update
• Refactoring (no functional changes)

Checklist

• I have read the Contributing Guide
• I have tested my changes locally
• My changes generate no new warnings
• Any dependent changes have been merged and published

Screenshots (if applicable)

Additional Notes

Built-in Providers (multi-model) Custom Providers (deployment)
├── Anthropic ├── Azure AI Foundry (user-created)
├── OpenAI ├── AWS Bedrock (user-created)
└── Google └── My Custom Endpoint (user-created)

Storage:
~/.openwork/
├── provider-configs.json ← Multi-config per provider
├── user-providers.json ← Custom provider definitions
└── .env ← Legacy (migrated on startup)

[Hunter Lovell (hntrl)]

Directionally I like the way that this is headed but there's a few nits to address:

• Can we pre-fetch/cache the available model strings for each provider when we first over the popover? There's a slight delay that comes from retrieving stored profiles from IPC
• We have the option to "name" a configuration for default providers, so my assumption is that we can select between multiple configs for that default. I don't believe this is the case though
• Selecting the pencil icon next to each model in the provider tab takes me to edit the API key config, but my assumption would be that's for configuring the model profile itself
• I like the idea of having model profiles to adjust things like model parameters and what not, but I think part of that would require upstreaming a way to get static chat model params to langchain (exporting a chat model zod schema, which is definitely something we could do). E.g. I can have a opus-super-omega-max model profile that maxes all reasoning and budgets out
• What are the custom providers you're looking to use? I think the UX for adding a provider that complies with a schema is a little odd right now. I'd expect something like:
  • you explicitly set a provider instead of an "api format"
  • if its an openAI provider, you can configure the base url in that dialog. Same if its an anthropic provider, etc.
  • Each of these would have its own "onboarding flow" to define provider-only auth methods or whatever we deem fit to put in there

For the lints + formats, I'm going to push changes to solve this on current main. If you wouldn't mind rebasing and descoping it to just the model configuration changes that would derisk the PR and be much appreciated 🙏