Issue 396 - Migrate to ClinVar XML V2

[apriltuesday]

Closes #396

• Parse XSD version from header so we can continue to parse historical V1 data
• Extract clinical significance info from the ClinVarRecord class into a new ClinicalClassification class
  • This is not really necessary right now, but enables us to model the additional structure more completely in the future.
• If there is a single clinical classification in an RCV and a single description for that classification, this is consistent with the current schema so we proceed as usual.
• If there are multiples, this is logged and counted as a skipped record, but will not crash the pipeline.
• Most of the other changed files are test datasets; for these I used the same RCVs but from a recent ClinVar release to keep things as stable as possible, but of course there are still differences...

[apriltuesday]

There's maybe another failure mode we're missing, which is if there's a single clinical classification but it's a string that isn't in this enum... this would include values like "Tier I - Strong" and others listed here.

At it stands this is only caught when the evidence string is generated and validated, because we don't maintain the list of acceptable values in this code base. I'm not sure if we should change this or just let these be counted as invalid and investigated.