feat: add remote storage loaders and writers (S3, GCS, Azure Blob, SQL)

[CoronRing]

Summary

• Adds S3Loader / S3Writer, GCSLoader / GCSWriter, AzureBlobLoader / AzureBlobWriter, and SQLLoader / SQLWriter under railtracks.loaders and railtracks.writers
• All providers use optional extras (railtracks[aws], railtracks[gcp], railtracks[azure-blob], railtracks[sql]) so the core package stays lean
• All loaders/writers expose sync and async interfaces (load/aload, write/awrite)
• SQL classes include a context-manager (with SQLLoader(...) as l) and explicit close() for engine lifecycle management
• SQL identifier arguments validated against a strict allowlist at construction time to prevent injection
• Full unit test coverage across all 8 classes (127 tests, all passing)
• Comprehensive developer docs under docs/integrations/storage/ with pip + uv install tabs, security callouts, and provider-specific auth guidance

Security hardening

• SQL table/column names validated at __init__ time — raises ValueError on any metacharacter ([A-Za-z_][A-Za-z0-9_$]* allowlist, supports schema.table)
• Helpful ValueError when content_column is missing from query results (was a bare KeyError)
• __repr__ on all classes exposes only non-sensitive fields (bucket/container name); credentials never appear in repr
• UserWarning emitted when prefix is passed to SQLLoader.load() or SQLWriter.write() (unsupported, silently ignored before)
• All ImportError messages include both pip install and uv add forms

Limitations documented

• CTE (WITH …) queries not supported as table_or_query; workaround shown in docs
• aload/awrite are thread-backed (asyncio.to_thread) not true-async; noted in docs with guidance for high-concurrency cases
• SQLWriter.write() is all-or-nothing (single transaction); partial-failure pattern documented

Test plan

• 127 unit tests passing across all 4 providers × loader + writer
• SQL tests use real in-memory SQLite (no mocks for correctness)
• Cloud tests (S3/GCS/Azure) use provider SDK mocks
• Async variants covered for all classes

🤖 Generated with Claude Code

[Pooria90]

Good work @CoronRing

I went over the code. A few substantial changes needs to be applied:

1. We need to move loaders and writers modules to packages/railtracks/src/railtracks/retrieval/loaders as they are being built on top of Amir's work on loaders. I recommend having all classes for a specific provider under one single module: for example having loaders/s3.py and writers/s3.py all under one s3.py.
   Or we can also have them under a loaders/cloud folder. We can discuss the structure.
2. The modules are returning Chunk which is located in our old vector_stores module. That module is deprecated now and will be removed from the framework. The new type that we use for our retrieval module is called Document which is located in packages/railtracks/src/railtracks/retrieval/models.py. Please refer to Amir's loaders for examples.
3. Please adjust the Base classes, data models, and the docs accordingly.