Fix _finishLongTextAscii returning negative remaining length

[hisener]

return len - outPtr was wrong when outPtr accumulated across multiple outer-loop iterations: len had already been decremented for previous iterations, so subtracting the full outPtr produced a negative value, causing _finishLongText to skip decoding the non-ASCII byte.

Fix: use return len - _inputPtr, which holds only the bytes consumed in the current iteration.

Fixes #686.

[cowtowncoder]

Ok, I think fix needs to against branch 2.21 since 2.18 and 2.21 are 2.x LTS versions and (2.18.x works).
I can merge forward from 2.x to 3.x but not vice versa.

[github-actions]

🧪 Code Coverage Report

Coverage Type	Coverage	Change
📝 Instructions		= 0.00%
🔀 Branches		= 0.00%

[cowtowncoder]

Aside from re-creating PR against 2.21, one process thing is that unless you have already sent it,
we'd need CLA (https://github.com/FasterXML/jackson/blob/main/CLA-jackson-2026.pdf).
This only needs to be sent once, and the usual way is to print it, fill & sign, scan/photo, email to cla at fasterxml dot com.

Once we have it, I can proceed with merging & CLA is good for all future contributions.

Thank you again -- looking forward to merging the fix!

[hisener]

Addressed the comment and rebased onto 2.21. Also sent the CLA email.

[cowtowncoder]

Thank you, @hisener ! Fix merged in, will be in 2.21.3 and 3.1.2.

[hisener]

@cowtowncoder Is there a planned date for the 2.21.3 release?

[cowtowncoder]

@hisener Depends on various things, including my availability (takes 2 hours of time to publish). Within next couple of weeks, maybe?

[hisener]

@cowtowncoder sounds good; thanks!