Use Locale.ROOT for case folding in CaseInsensitiveNameSet

[Sahana2524]

CaseInsensitiveNameSet folds names with the JVM default locale, so case-insensitive element matching quietly breaks for names containing I/i once the default is a locale like Turkish:

• construct() and contains() both call String.toLowerCase() with no locale, so a wrapper name like ITEM stops matching an incoming item when the default locale is tr ("ITEM".toLowerCase() becomes dotless ıtem)
• the names being matched are element/attribute names from the parsed document, and this feeds unwrapped-list wrapping via FromXmlParser.addVirtualWrapping when ACCEPT_CASE_INSENSITIVE_PROPERTIES is on, so the result ends up depending on server locale
• XmlTokenBuffer already matches the same names with equalsIgnoreCase, which is locale independent, so this helper is the odd one out

Both calls now use Locale.ROOT. Pure-ASCII names, the common case, are unchanged.

[github-actions]

🧪 Code Coverage Report

Metric	Coverage	Change
Instructions	73.55%	📈 +0.020%
Branches	68.08%	📈 +0.050%

Coverage data generated from JaCoCo test results

[cowtowncoder]

Looks good -- just need CLA.

[Sahana2524]

Sure, I'll email the signed CLA to info@fasterxml. Saw your note on #878 that one copy covers both PRs, so that's the plan.