do not merge: bananza mode wip

[aaronsteers]

Summary by CodeRabbit

• New Features

  • Added lake storage integration enabling fast unload/load between caches and cloud object storage.
  • Introduced S3 and GCS lake backends with a default path convention; supports per-stream and bulk operations.
  • Enabled fast copy workflows for Snowflake and BigQuery caches to and from lake storage.

• Documentation

  • Added an example script showcasing an end-to-end Snowflake → S3 → Snowflake data transfer with performance summaries.

• Chores

  • Exposed lake storage classes at the top-level package for simpler imports.