v0.4.0 CSV mode - API key is optional from now on

Added

• Enhanced provider information retrieval:
  • New get_provider_info() method in both API and CSV modes
  • Combines live provider data with downloaded dataset information
  • Automatic fallback to CSV catalog on API errors
  • Comprehensive metadata about downloaded datasets
• CSV catalog fallback functionality (#5):
  • Automatic fallback to CSV catalog when API key is not available
  • Optional force_csv_mode to always use CSV catalog
  • Lazy initialization of CSV catalog to improve performance
  • ID normalization for consistent provider lookup
• Improved error handling:
  • Graceful handling of 413 (Request Entity Too Large) errors
  • Automatic fallback to CSV mode on authentication errors
  • Better token handling with empty headers when no token is available
  • Robust handling of CSV network errors with fallback to cached data
  • Proper handling of empty metadata files during concurrent access

Changed

• Made CSV catalog initialization lazy to reduce unnecessary directory creation
• Improved token handling to return None instead of raising errors
• Enhanced error messages and logging for better debugging
• Improved file locking mechanism:
  • Fixed handling of file modes for concurrent access
  • Added proper file existence checks before opening
  • Enhanced empty file handling during metadata operations

Developer Changes

• Added comprehensive tests for provider information retrieval
• Added tests for CSV catalog functionality
• Added tests for ID normalization and conversion
• Added tests for lazy initialization behavior
• Added tests for CSV network error scenarios
• Added tests for concurrent file access edge cases
• Updated documentation with CSV fallback examples and API reference

Full Changelog: v0.3.0...v0.4.0