Most of the classes and predicates in this query can be shared between
the two languages. There's just a few language-specific things that we
place in IncompleteMultiCharacterSanitizationSpecific.
rb/incomplete-sanitization has a few cases where we find flow from one
one string substitution call to another, e.g.
a.sub(...).sub(...)
But this didn't find typical chained uses of the destructive variants,
e.g.
a.sub!(...)
a.sub!(...)
We now handle those cases by tracking flow from the post-update node for
the receiver of the first call.
