feat(code): add function to download notebook outputs

[KeijiBranshi]

BUG=b/371574828
CHILD=#185
BLOCKED_BY=go/kaggle-pr/32581,go/kaggle-pr/32632
CC=@rosbo,@dster2,@jplotts

Extends the same functionality from models.py, datasets.py and competition.py to the notebooks at https://kaggle.com/code.

Changes

handle.py

• added a new *Handle data type
• using Code to align with the route on the main site https://kaggle.com/code

cache.py

• added new functions to dictate the cached path for notebook outputs based on the properties in CodeHandle
• this mostly mirrors the same structure as the model, dataset, and competition paths.
  • the cache structure is split into the output_path, archive_path, a _completion_file_marker_path for individual files in the download payload, and a _completion_file_marker_path for the entire download payload.
• the structure is as follows:

  <cache_root>/
  └── notebooks/
      └── username/
          └── notebook_slug/
              ├── output.complete  <-- tracker for the entire output
              ├── .complete/       <-- trackers for per file within the output
              │   └── output/
              │       ├── file1.txt.complete
              │       ├── file2.txt.complete
              ├── output.archive   <-- the compressed output (.tar.gz or .zip)
              └── output/          <-- the uncompressed output
                  ├── file1.txt
                  └── file2.txt
  

http_resolver.py

• Implemented the NotebookOutputHttpResolver
• Note, we don't currently have an API endpoint to download notebook output in a kagglehub-compatible compression format (left a TODO with our internal tracker for this)
• It leverages the our existing KaggleApiV1Client + the new cache location mentioned above

registry.py + __init__.py

• bootstraps the NotebookOutputHttpResolver so that it can be called by kagglehub.notebook_output_download in code.py

code.py

• the entry point to the notebook output downloading functionality
• the file is named code.py to align with our navigation paths at https://kaggle.com (similar to models, datasets, and competitions). Open to changing if needed.
• the function is named notebook_output_download to be more specific about what's being downoaded

test_notebook_output_download.py

• integration tests for the new kagglehub.notebook_output_download function
• TODO([DRAFT] test(code): add integration tests for code.notebook_output_download #185): adding tests in a followup since that requires propping up a stubbed API server. Trying to keep this diff from getting any bigger.

gcs_upload.py

• A miscellaneous lint error that slipped through. Fixing here as a drive-by change as per this comment

[KeijiBranshi]

note: these are miscellaneous drive-by changes as per #issuecomment-2494710160

[KeijiBranshi]

fyi @calderjo going to merge now to unblock. lmk if you have any comments though and I can address in a follow up iteration