feat: add CSV, JSON, and Parquet URL data providers

[MDUYN]

Summary

Add support for loading external CSV, JSON, and Parquet data from any URL during strategy execution.

Changes

• BaseURLDataProvider — shared base class with caching (in-memory + file), refresh intervals, date parsing, and pre/post-processing callbacks
• CSVURLDataProvider, JSONURLDataProvider, ParquetURLDataProvider — format-specific implementations
• DataSource.from_csv(), from_json(), from_parquet() — factory methods for pre-declared data sources
• context.fetch_csv(), fetch_json(), fetch_parquet() — on-demand methods for dynamic fetching
• Cache path fix — cache files now stored inside resources/data/ (synced by state handlers in cloud deployments) instead of a detached .data_cache/ directory
• 43 tests covering all three providers
• Documentation — new external-data.md docs page, sidebar registration, README updates

Cloud deployment support

Cache files live inside the resource directory, so AWSS3StorageStateHandler and AzureBlobStorageStateHandler automatically persist them. refresh_interval works correctly across cold starts using file modification times.

Closes #458