Optimize instruction size

[gilescope]

If you look in the std lib we do these tricks in to_digit.

Smaller wasm is a good thing... (godbolt: https://godbolt.org/z/vocdaKW9x )

I know speed is not the prime reason for the crate, but bench showed a 20% speedup of hex::encode when I ran them.
(The rest of the benches looked no different.)

[aurexav]

Smaller wasm is a good thing... (godbolt: godbolt.org/z/vocdaKW9x )

Hey! Thanks for the PR!

I found this also makes the instructions smaller.
Same instructions, same speed but more readable.

fn is_hex_ascii(byte: &u8) -> bool {
	// Convert to lowercase.
	let byte = byte | 0b10_0000;
	
	matches!(byte, b'0'..=b'9' | b'a'..=b'f')
}

[gilescope]

Short and sweet, I like it.