Determine whether we should handle off transcript c.HGVS

Consider an upstream HGVS such as "NM_000133.4(F9):c.-53A>G" (-26)

The HGVS guidelines say it's upstream and shouldn't use the transcript to classify. However, people do report it that way.

* VEP will generate the (incorrect) c.HGVS
* Alissa will accept this coordinate (tested today)
* Mutalyzer will follow the HGVS spec and report it as an error
* Shariant will currently error importing that as it's off the transcript

We should determine whether we want to accept these.

VG can correctly resolve it, if we allow an option to disable the transcript bounds checking

It is unknown whether it will load into ClinVar