ETT-1459: reuse main_repo_audit for extracting info from mets

[aelkiss]

• rename to "crawl repo mets" and remove md5 checking (done by truenas_audit.pl)
• record date of first ingest to feed_audit; extracts all PREMIS 'ingestion' events & takes the first
• separate flag for source mets handling
• extract source METS PREMIS events
• extract methods to make testable (similar to treatment for populate_rights)

This is the second half of ETT-1459 -- backfilling feed_audit with the first ingest date for things that don't have it.

This used the same approach to getting it under test as I did for populate_rights_data.pl -- I tried calling out with system but ran into some of the same issues you did with Glacier with not inheriting the config. I figured making it more directly testable was probably preferable, and I think it's close to a point where it could be changed to be an object to further make it testable.

Of particular note is that some of the things in crawl_repo_mets are still duplicative of truenas_audit -- probably we should extract those somewhere separate or do less rigorous checks in crawl_repo_mets.

I don't love that the interface is outputting to stdout, but I can redirect the output to a file in Kubernetes and process later.

[moseshll]

Yes, lotsa stuff we've seen before. Could see if HTFeed::RepositoryIterator would do some of the lifting, maybe at a later point. It would probably have some gotchas to refactor since it's really tailored to the truenas audit -- not worth the pain right now IMHO.