Two main features were added.

One, various sequence closers can imply a subgroup. This is mainly useful
in contexts like arrays, where you want one subgroup per array element.

Two, predicated opening/closing delimiters. These allow having both an
infinite set of opening delimiters, as well as having a set of closers
that is both infinite and dependant on what the opener was.

Note, however, that predicated openers and closers will be slower than
using normal openers/closers, since every one has to be tried to check if
a token is a opener/closer.

This fixes subgroups, by mostly rewriting the way the delimiter works.

Moved the parsing utilities SequenceDelimiter and TokenSplitter to the
parserutils package, instead of the funcutils package.