figure out a way to make it even slightly faster

[cabfile]

the worker script feels unoptimized (maybe even because it is?)
i can think of a few potential optimizations but its possible they wont even make a difference
still worth a try imo

[cabfile]

with commit 623490e, its probably 2-5% faster now (though i could be wrong)
still think its possible to push it further

[cabfile]

hmm. what if cabbr could straight up compile c expressions whenever possible, and then just read their output data? this way rendering them could become way faster