Reject a private-key hex string with a non-hex digit

[SashaMIT]

Summary

• hexToBytes throws when the length is odd, and it accepts any other character.
• hexToBytes("0G") returned. G is not a hex digit. The nibble math maps it to 16, so 0G decodes to the same byte as 10.
• A non-hex digit now throws BitPayGenericException. 0123456789abcdef still decodes.

Test plan

• On tip, it_should_reject_a_non_hex_key_digit expected BitPayGenericException and nothing was thrown
• After the change, mvn -Dtest=KeyUtilsTest test is 6 tests, 0 failures

Made with Cursor

[aharoitx]

Thanks for the fix, Sasha.

I checked it: on the current code "0G" decodes to the same byte as "10", so a bad key turns into a different key without any error. Your test fails on 10.3.x and passes with the change. It also matches what the C# SDK already does.

Tests and checkstyle pass on Java 8. I'll fix the small import conflict with 10.3.x and merge after CI passes.

[SashaMIT]

Thanks for checking it on 10.3.x and for merging it. A bad hex digit was becoming a different key, and rejecting it matches the C# SDK.