Because the point of Evaluatable things is that this static/dynamic distinction isn't needed.

But for LengthEv, even if it is a constant, if the encoding is variable-width the length might not be static. Darn UTF-8 is such a pain in the butt.

Add a comment that this is why we need this split, and that LengthEv is not sufficient by itself.

That is, assuming my reasoning here is correct.

But also, consider the comment in ElementBaseGrammarMixin.scala. Maybe we just need EffectiveLengthEv and then we don't need this static/runtime split in the combinators/primitives?