Use SDK parallel ranged downloads for large file saves

[goanpeca]

Summary

The SDK includes parallel/ranged download primitives that can make large file saves faster and more reliable. The VS Code UI should use those paths for large downloads instead of treating every save as a simple single stream.

SDK capability

• B2Object.createReadStream(fileId, totalSize, { rangeSize, concurrency, signal })
• createParallelDownloadStream(raw, accountInfo, { fileId, totalSize, rangeSize, concurrency, maxRetries, signal }) from @backblaze-labs/b2-sdk/download/parallel
• Bucket.download(..., { range, onProgress, signal })

Proposed UI

For Save As / download-to-folder flows, switch large files to the SDK's parallel ranged download path with progress and cancellation. Keep smaller downloads on the simple path if that is simpler and faster.

Acceptance criteria

• Large file downloads use ranged/parallel SDK primitives.
• Users see accurate progress during the full download.
• Cancellation aborts in-flight ranged requests.
• Concurrency/range size are bounded and configurable internally or through settings.
• Tests cover large-file download orchestration and cancellation behavior.

Related

• Download files to a chosen location (Save As / download to folder) #57 tracks adding general Save As / download UI.