feat: implement DeepsetClient for API communication

[mathislucka]

Overview

This PR implements a DeepsetClient class to handle all API communication with the deepset Cloud API. The client provides a more structured, testable, and maintainable approach to API interactions.

Changes

1. Created a new DeepsetClient class with the following features:

   • Async implementation with context manager support (async with DeepsetClient() as client)
   • Dependency injection for testing via the HttpClient protocol
   • Consistent error handling and response processing
   • Methods that map to API endpoints used in the application

2. Updated main.py to use the new client:

   • Added helper function to run async code in sync context (async_to_sync decorator)
   • Migrated all API request functions to use the new client
   • Updated tool functions to be async and use the client

3. Added tests:

   • Unit tests for the DeepsetClient class
   • Updated existing tests to work with the async client

4. Updated dependencies:

   • Added httpx for modern async HTTP requests
   • Added typing-extensions for Python 3.11 compatibility with Self

Implementation Details

• The DeepsetClient implements the AsyncContextManager protocol to properly handle client resources
• A thin HTTP client abstraction enables dependency injection for testing
• All original functionality is preserved with improved testability
• Error handling is consistent across all endpoints

Testing Considerations

• Unit tests verify the client can be used with environment variables or explicit parameters
• Tests check error handling for missing credentials
• Mocks verify that requests are made with correct endpoints and parameters

Next Steps

Future PRs could enhance this implementation by:

1. Adding specific models for API responses
2. Creating resource-specific classes like PipelineResource
3. Introducing proper API exception hierarchy
4. Adding more comprehensive error handling and retries

Closes #10