Fix != to return true when either operand is NaN

[SABITHSAHEB]

Thanks for your contribution to Apache Commons!

• Read the contribution guidelines for this project.
• Read the ASF Generative Tooling Guidance if you use Artificial Intelligence (AI).
• I used AI to create any part of, or all of, this pull request. Which AI tool was used to create this pull request, and to what extent did it contribute? Claude Code found the defect and wrote the fix and the test updates; I reviewed the change before submitting.
• Run a successful build using the default Maven goal with mvn; that's mvn on the command line by itself.
• Write unit tests that match behavioral changes, where the tests fail if the changes to the runtime are not applied. This may not always be possible, but it is a best practice.
• Write a pull request description that is detailed enough to understand what the pull request does, how, and why.
• Each commit in the pull request should have a meaningful subject line and body. Note that a maintainer may squash commits during the merge process.

CoreOperationCompare.equal, the shared evaluator for = and !=, returns false early when either operand converts to NaN, which bypasses the result ^ invert step that turns the equality result into !=. XPath 1.0 section 3.4 compares numbers per IEEE 754, where NaN is not equal to anything, including itself, so $nan = $nan is false but $nan != $nan, $nan != 0 and number('foo') != 1 must all be true; Xalan and libxml2 both answer true. JXPath answers false for both operators, so a predicate like txns[amount != 0] silently drops a node whose caller-supplied amount is a non-numeric string (its number value is NaN) instead of selecting it, and a filter written to catch every value different from the expected one is bypassed by a value that does not parse as a number.

This completes JXPATH-95: the reporter pointed out there that $nan != 1 must be true, but the follow-up commit 40689aa made every NaN comparison return false and encoded that in testNan. Java's == on doubles already follows IEEE 754, so the fix deletes the early returns and lets the comparison flow through the inversion. =, <, >, <=, >= with NaN all stay false, and empty node-set comparisons are unaffected (testEmptyNodeSetOperations still passes). The updated testNan expectations and the new string-coercion assertions fail on the current code and pass with the fix; the default mvn goal passes.