Consider alternative method of checking if a character is malformed

[jakewilliami]

We do this here, but this check is presumably computationally expensive, as we do bitwise reinterpretations and counting of leading ones and zeros, etc.

Extends #1

[jakewilliami]

Do we even need this ismalformed check? We only care about code points between 12 and 18. Do we care if we can correctly classify them as 30 or 31? Not sure