Reject truncated percent-encoding in RFC2231Utils.decodeText

[alhudz]

Repro: parse a Content-Disposition value of form-data; name="f"; filename*=UTF-8''report%3 (a truncated RFC 5987 percent escape) and read the resulting file name.
Expected: rejected, the same as %GG or a non-ASCII byte already are.
Actual: RFC2231Utils.decodeText returns report; fromHex treats a % with fewer than two following characters as end-of-input and returns what it decoded so far, dropping the truncated escape.
Fix: throw IllegalArgumentException on a truncated % escape, so the value is rejected like every other malformed input in fromHex and like the sibling QuotedPrintableDecoder. Covered by a new case in RFC2231UtilityTestCase.

[garydgregory]

Thank you @alhudz , merged 🚀 Please port to the 1.x branch.

[alhudz]

Thanks for merging. Port to 1.x is up as #477, same change adapted to RFC2231Utility, full suite green there.