use new Isovar API but keep old logic for determining variant effects

[iskandr]

This PR is an attempt to make the smallest possible change while still bringing Vaxrank over to the new Isovar API. To do this I'm leaving some redundant code in Vaxrank for e.g. determine variant effects, which can be removed in a later PR.

Also:

• cleaning up CLI code by pulling out little helper functions
• created run_vaxrank_from_parsed_args, which simplifies testing
• moved logic for --version into ArgumentParser instance

Importantly this PR allows use of newer versions of pysam (>0.9), as requested in https://github.com/openvax/vaxrank/issues/179 and by several other people offline

After the next PR I'll start actually re-running PGV patients and comparing reports.

[coveralls]

Coverage increased (+0.2%) to 88.2% when pulling 3b0be11c66cef9bd6d400fab1fba0d91a2e0d6cf on use-new-isovar-api-but-old-effects-logic into d0e74d976bd5e493b259a6b2fbee204f195361dd on master.