Add encoding field to pstring (and cstring?)

[roblabla]

Currently, most protodef implementation assume pstring is utf-8 or ASCII. Fortunately, that's compatible with a lot of protocols. But this is a big shortcoming.

It would be a good idea to allow the protocol author to write what encoding a pstring is in. I wonder if there's a standard about encoding names somewhere ? A quick look gave me https://encoding.spec.whatwg.org/#names-and-labels .

pstring's count argument should count the byte length, regardless of the encoding. I'm fairly sure this is how most protocol send strings.

cstring has a similar problem. We should limit it only to single-byte encodings, given it is meant to be compatible with C char*.