Invalid hex chars

[xtonik]

Added missing check for digit characters being equal to '0' or greater. As the used way of tightly coupled SWAR testing of character being is around byte range, some character from upper half of ASCII were considered as valid also not filter by existing checks and thus check for characters being less or equal to '9' added also.

There may be better way - smaller number of operations - for testing hexadecimal characters after this change.

There were some troubles with little / big endian integer encoding, it should be verified whether they do not occur on other platform than tested - Intel.

[wrandelshofer]

Thank you very much for providing a fix for this issue. 😀
Based on your merge request, I have now commited a fix that uses one operation less. I made multiple commits to the main branch. The most significand was probably this one: a67caba