Add Unicode character name support and fix bugs in tr/// operator

[fglock]

Summary

This PR improves the transliteration operator (tr/// and y///) by adding Unicode character name support and fixing critical bugs.

Test Results

• Before: 256/318 passing (80.5%) - test died at line 1113
• After: 277/318 passing (87.1%)
• Improvement: +21 tests (+6.6%)

Changes

1. ✅ Unicode Character Name Support

• Added support for \N{name} syntax with actual Unicode character names
• Uses UnicodeResolver integration with ICU4J (~30,000+ names supported)
• Properly rejects multi-character named sequences

Example:

$s = "é";
$s =~ tr/\N{LATIN SMALL LETTER E WITH ACUTE}/E/;  # Now works!

2. ✅ Empty \N{} Validation

• Added validation for empty character names
• Now gives proper error: Unknown charname ''

3. ✅ Surrogate Pair Bug Fix

• Fixed incorrect handling where \x{ffff} was being removed
• Changed from checking isHighSurrogate() to isSupplementaryCodePoint()
• Now correctly preserves standalone surrogates and valid surrogate pairs

Example:

no warnings 'utf8';
$s = "\x{d800}\x{ffff}";
$s =~ tr/\0/A/;
# Before: "\x{d800}" (lost \x{ffff})
# After: "\x{d800}\x{ffff}" (preserved both)

Files Modified

• src/main/java/org/perlonjava/operators/RuntimeTransliterate.java
• src/main/java/org/perlonjava/regex/UnicodeResolver.java

Documentation

• TR_UNICODE_NAME_SUPPORT.md - Technical details
• TR_IMPROVEMENTS_SUMMARY.md - Complete summary

All changes maintain Perl 5.36 compatibility.

[fglock]

Closing this PR as requested. The tr/// improvements (+21 tests) are solid, but there are concerns about test regressions. Note: Investigation showed the listed regressions (magic.t, avhv.t, etc.) were introduced by PR #62 (strict-refs fix), not by these tr/// changes. The work is preserved in the feature/tr-unicode-support branch for future reference.