Build info/version info inside ADAM-generated files

[nealsid]

We should build off of Sebastian's work in #138 to output ADAM version info inside files generated by ADAM, so that we can version files containing ADAMRecords, ADAMNucleotideFragments, ADAMVariants, etc.

[nealsid]

Also, I may have missed some previous discussions on how we do this, but I recently converted hg19 to a Parquet file of ADAMNucleotideConfigFragments. It seems there's no way to recover the reference version information - or am I missing something? The AVRO record contig fields don't store this. Can we shove it in the Parquet metadata somewhere?

[tdanford]

Calling out @massie here (when you get back from vacation, Matt) -- he's had some thoughts on embedding information into the Parquet metadata.

[fnothaft]

Ping @massie

[massie]

Once we upgrade to Parquet 1.6.0, we'll be able to read/write arbitrary metadata much more easily. We can easily drop the version info (introduced in #138) into the metadata to help with debugging.

The upgrade to 1.6.0 is going well but three tests are failing because of issues with predicates (UnboundRecordFilter).

[heuermh]

Is this worth another look? Parquet dependency is now at version 1.8.x.

[fnothaft]

Perhaps we can write this with our various metadata?

[fnothaft]

We should resolve this as part of #1257.

[fnothaft]

This will be resolved as part of #1257. Closing as dupe.