Fix UTF-16 surrogate validation

[efegokdemir]

Summary

Fixes #287 by correcting UTF-16 surrogate validation in UTF8Util.isWellFormed.

Problem

The previous logic rejected valid high/low surrogate pairs and accepted unpaired surrogates, including a trailing high surrogate.

Fix

• Accept valid surrogate pairs.
• Reject an unpaired low surrogate at the start or in the sequence.
• Reject a high surrogate not followed by a low surrogate.
• Reject a trailing high surrogate after the loop.
• Preserve the existing null-character and sanity checks.

Regression tests

Updated UTF8UtilTest.mustNotChars to cover:

• a valid surrogate pair;
• an unpaired high surrogate;
• an unpaired low surrogate;
• a trailing high surrogate.

Testing

• ./mvnw -pl bifromq-util -am -DskipTests=false test — attempted but blocked: the environment has no Java runtime.
• git diff --check — passed.

Closes #287.

Contribution

AI assistance was used during investigation and implementation; the change and validation results were reviewed before submission.