Reorganize file_level_rules.json

[rwblair]

Current implementation has lots of repeated regex. I was thinking along the lines of putting all the entities and their regex in an object. Then when defining valid file names we can just have a list of entities key names (and if they are required or optional) instead of the regex itself.

I think we could also provide better feed back on filename issues if we parsed the file extension first and selectively chose which rules to apply, and if it fails return information on whats required and optional. Doesn't look like there's a way to get the regex show you points of failure. One option would be to apply entity regex consecutively against a file name, and consume the entities as they match. This way we could find specifically which ones fail.

[rwblair]

We should use these schema files to enforce required entities: bids-standard/bids-specification#475