Add fsm_intersect_charset(), fsm -U

[katef]

Add fsm_intersect_charset(), a convenience to intersect an fsm against a given character set. I pulled this out of the forthcoming rx(1) tool.

Unlike fsm_intersect(), this doesn't automatically determinise its operands. I'm not convinced fsm_intersect() should do that, either. In most cases we care enough about performance to do this (or rather to avoid doing it unnecessarily) in the caller.

I've exposed this as fsm -U <charset>:

[silentbicycle]

Looks good to me.

Is it worth noting that the isn't any interpretation done to the character set? e.g. a-z is treated as ['a', '-', 'z'] rather than [a-z]? Or adding special handling if the character set starts with [ and ends with ]? That seems like it would be a common use case.

[katef]

I removed group parsing as an entry point into a regex grammar in 2ede0ed, which was exposed as a similar cli argument for re(1). I think I'd either want to reintroduce that (somehow...), or invent a tr(1)-style syntax (perhaps as its own dialect?) just for this. But I wasn't ready to do that right now.