option to use 'deep' caching for small files, metadata-only caching for large ones

[notestaff]

TL;DR: add option to use 'deep' caching for files below certain size, normal metadata caching for larger files.

Git-versioned files often get their metadata updated (when switching branches) without changing the content. There are other files (e.g. config files generated by earlier pipeline steps) where, again, file metadata change should not trigger recomputation, because the content is the same. Currently, avoiding recomputation requires turning on 'deep' caching, but that's infeasible if some input files are large. Some code in nextflow.util.CacheHelper tries to infer from heuristics whether deep caching should be used ("if it's an asset in the repo it's likely small"), but using the actual file size would be better. A global parameter could control the size below which files' content would get hashed.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.