I noticed that a couple of usages of PL_MIN(PL_MAX and vice versa
involve a wasteful double evaluation. Plus, the argument order of
PL_CLAMP is way easier to remember.