Background

• RFC: https://www.notion.so/325c3a490d988106badbf24255ba38e1
• Blocked by: https://github.com/rwx-cloud/cloud/pull/7104

Server-side RFC-166 standardizes log downloads to always return zip archives with a run_id field. This PR is the CLI side of that change (RWX-164).

Problem

The CLI had dual GET/POST logic and optional extraction (--auto-extract). This made it impossible to deliver background process and ready-check logs alongside the main task log — a zip archive is the natural container for multiple files, but the old behavior only extracted when explicitly requested.

With RFC-166, the server always returns a zip with a run_id, so the dual path is unnecessary and the default behavior should always extract into a structured directory, making background process logs available automatically.

Solution

• Default behavior: always extracts the zip into <output-dir>/<run-id>/ and prints the directory path + each extracted file
• --zip flag: new flag to skip extraction and save the raw zip archive instead
• --auto-extract: retained as a hidden no-op flag for backward compatibility
• API layer: removed the GET code path from DownloadLogs; now always POSTs {token, filename, contents} as form data
• LogDownloadRequestResult: Contents changed from *string → string; added RunID string
• Tests updated throughout to reflect new behavior

╰─⠠⠵ RWX_HOST=cloud.local ./rwx logs a419fbbbee774d878d782bbb431e290c --task input
Logs downloaded to /Users/jasonrobinaugh/code/cli/.rwx/downloads/a419fbbbee774d878d782bbb431e290c/
  /Users/jasonrobinaugh/code/cli/.rwx/downloads/a419fbbbee774d878d782bbb431e290c/input.log
  /Users/jasonrobinaugh/code/cli/.rwx/downloads/a419fbbbee774d878d782bbb431e290c/input.bg.sleep.log

Further confirmation needed

• Smoke test once rwx-cloud/cloud#7104 is merged and deployed: rwx logs <task-id> should extract to .rwx/downloads/<run-id>/, and rwx logs <task-id> --zip should save the raw archive