Add support for switching RLE subunit between largest and smallest

[theferrit32]

The code right now uses the largest subunit obtained after trimming common sequence.

The goal of this issue is to add support for instead using the smallest. This will then be made the default when we update to a forthcoming spec version, and will enable us to support creating digests using the prior 'largest subunit' method optionally in the code, when that is requested.

vrs-python/src/ga4gh/vrs/models.py

Line 688 in a50b9fb

def ga4gh_serialize_as_version(self, as_version: PrevVrsVersion) -> str:

vrs-python/src/ga4gh/vrs/normalize.py

Line 207 in a50b9fb

factors = _factor_gen(seed_length)

See #549 (comment) for prior discussion.