Improve try_parse and encode performance

[Nugine]

I'm submitting a refactor

Description

This PR improves the performance of Uuid::{try_parse, parse_str} and {Simple, Hyphenated, Braced, Urn}::{encode_lower, encode_upper}

bench	before	after
encode_hyphen	25 ns	15 ns
encode_simple	23 ns	15 ns
encode_urn	29 ns	16 ns
hyphenated	43 ns	37 ns
simple	39 ns	32 ns
urn	46 ns	37 ns
parse_invalid_character	72 ns	63 ns
parse_invalid_group_len	81 ns	73 ns
parse_invalid_groups	94 ns	90 ns
parse_invalid_len	88 ns	86 ns
parse_nil	46 ns	26 ns
parse_nil_hyphenated	47 ns	28 ns
parse_random	46 ns	26 ns
parse_random_hyphenated	47 ns	28 ns
parse_urn	49 ns	29 ns

Related issues

#551
#561

[KodrAus]

Maybe we could move the SAFETY comment to outside of the unsafe block here to also cover the ptr::write call?

The as_mut_ptr call isn't unsafe so we could also pull that one out.

[KodrAus]

Thanks for working on this @Nugine! I just had one nit-picky comment around SAFETY comments on the unsafe blocks.

It would be great to come back through and add some docs describing how we parse the various shaped UUIDs. At the moment I think a casual passer-by might have a hard time following along. I'm happy to come back and do that later though unless you'd like to write it up.

[KodrAus]

Maybe we could move the SAFETY comment to outside of the unsafe block here to also cover the ptr::write call?

The as_mut_ptr call isn't unsafe so we could also pull that one out.