Create hierarchical dates for opengeometadata records

[thatbudakguy]

Closes #1407

Tested by using it to reindex OGM records in stage, where you can currently see the result:

I made one or two very small changes from the traject implementation; notably other people aren't very good at providing their index years as integers...

[jcoyne]

Do we have any BCE dates we need to test? Any non-Gregorian calendar dates? Would it make sense to limit the top end of the range to the max year in the corpus? Can you expand one of the years to show what it looks like unfurled?

[thatbudakguy]

I've just reindexed all of prod and all of the OGM data and there are no BCE dates. If there are non-gregorian dates, I guess we have no way of knowing, because none of them are 4-digit integers that are obviously outside the range of gregorian dates...but I think that'd be at least somewhat of an abuse of the spec, if not outright invalid.

Re: limiting the top end to the max year, that seems like a nice change, although I'm not sure offhand how to do it and it might make the facet logic a lot more complicated.

You can see what they look like unfurled in current production, since the reindex is done.