tr: fix complemented class truncation ordering

[can1357]

uutils tr applies the complemented-character-class guard before --truncate-set1, then truncates the translated set afterward. That allows -ct to translate only a subset of complemented bytes, while GNU rejects this shape instead of allowing partial mapping.

Reproduction Steps

printf A | tr -ct '[:digit:]' X
# Expected (GNU): exit 1 with "when translating with complemented character classes, string2 must map all characters in the domain to one"
# Actual (uutils): exit 0 & outputs "A"

Impact

This breaks GNU compatibility and weakens sanitization pipelines that rely on tr -ct to enforce translation coverage. Inputs that should be rejected can be accepted with under-filtered output.

[cakebaker]

This test is incorrect because it contradicts your reproduction steps:

printf A | tr -ct '[:digit:]' X
# Expected (GNU): exit 1 with "when translating with complemented character classes, string2 must map all characters in the domain to one"

[can1357]

my bad, it's the rejection test. will push the inverse

[sylvestre]

@can1357 did you fix it? thanks

[github-actions]

GNU testsuite comparison:

Skipping an intermittent issue tests/cut/bounded-memory (passes in this run but fails in the 'main' branch)