feat: cache parsing results of current dependency files

[maxrake]

This change speeds up overall execution time, especially for manifests.
It does this by caching the results of parsing dependency files from the
current (HEAD) location under analysis. This reduces the instances of
parsing the same file from three to two. Previously, it would happen:

• When filtering files supplied as input
• When calculating the new dependencies
  • File has changed (or --force-analysis provided) and --all-deps
    not supplied
  • This one is eliminated via caching
• When submitting the files for Phylum analysis
  • During the PhylumApi.parseLockfile extension API call
  • This one remains since the extension API does not recognize the
    Python cache

There is also large change to rename internal uses of lockfile term.
This was done in a separate commit and includes the following basic
changes:

• Rename the phylum.ci.lockfile module to phylum.ci.depfile
• Rename the Lockfile class to Depfile
• Change internal naming uses of "lockfile" to "depfile"
• Change text/comment uses of "lockfile" to "dependency file"
• Keep the instances where Phylum CLI still uses "lockfile" naming
  • Related names are similarly unchanged, to stay consistent
  • These may be changed if/when the CLI changes to "depfile" language
• Keep the --lockfile argument for the phylum-ci command
  • Changing this would be a breaking change
  • It would need to happen in coordination with documentation updates
  • It may happen separately at a later time

Testing

Changes from this PR have been built in Docker images and hosted on Docker Hub for my account, as the cache_current_parsing tag.

Testing was completed with the private isildurs_bane repo, where it saved ~40s (2m40s --> 2m).
It was also completed with the public deepcausality-rs/deep_causality repo, where it saved ~3s (23s --> 20s).

[cd-work]

import cache 😄