Remove unnecessary call to to_ascii_lowercase

[tcharding]

We already iterate over lowercase characters when encoding a bech32 string, no need to call to_ascii_lowercase.

Add a unit test as documentation of this behavior as well as to verify the change is correct.