Create a performant, cloud-agnostic way to download & upload files to cloud buckets.

See discussion here: https://github.com/google/weather-tools/pull/254#discussion_r996220532

To investigate:
- Can we do better than `shutil.copyfileobj`?
- What are the optimal chunk sizes?
- Can we copy data in parallel?
- Are there optimizations we can do for large files?

One idea that @bahmandar has explored is calling gsutils in a subprocess (the CLI is really efficient at file transfer).