Crawl extension

[gadomski]

Over at https://github.com/gadomski/heystac I've implemented rudimentary crawling:

• Visit each child catalog and collection
• For each collection, fetch one item either from its item links or from its items endpoint

This seems generalizable to an extension, since the output could be useful it if were structured (things like crawl:start_datetime and crawl:end_datetime, roundtrip request times, etc). We should implement in stac-extensions, write up a spec and a json-schema for it, and then maybe make it its own https://github.com/stac-extensions.