Flexible backend class for metadata from file names and headers

[pbeaucage]

A lot of loaders need to ingest metadata from file names or text header files…having a generic/abstract class for this would be good. Ideally the input would be a regex and a list of names to extract from that regex matcher. @martintb suggested that the base Python regex might provide a parser that just does both these functions, but integrating into the OO tree of PyHyper is going to be a bit hairy.