Simplify sorts and review parsing in KMIR-AST

[jberthold]

A few improvements to make working with the KMIR-AST::Pgm easier.

• The AllocBytes sort in body.md contains a sequence of optional byte values. This seems to be an unnecessary complication.
• The ProvenanceMap is currently parsed as a simple Int list but could be more structured.
• A few other FIXME notes from prior work on the python parser can be removed now

[jberthold]

As it turned out from the integration tests we have, the bytes in constant operands may not always be populated in stable-mir, i.e. there can be null values in the byte sequence.
From inspection of the tests that fail on this, it appears as though the nulls' intention is to reserve space/pad memory for an optional value that is None, the None being represented as [0,0,0,0,0,0,0,0] and a sequence of 8 nulls following as padding for the size of the type that could be here.
Parsing the bytes values will be adapted to use zero for any null values.