[common] Fix CHAR min/max statistics collection

[beetle0915]

Purpose

Linked issue: close #4066

DataTypeChecks.isSupportedStatisticsType() lists CHAR as a supported
statistics type, but LogRecordBatchStatisticsCollector does not handle
CHAR when collecting min/max values. As a result, the bounds remain null
and predicates on CHAR columns cannot prune record batches.

LogRecordBatchStatisticsWriter also does not handle non-null CHAR
min/max values, so adding collector support alone would cause statistics
serialization to fail.

Brief change log

• Add CHAR min/max collection using InternalRow#getChar,
  BinaryString::compareTo, and BinaryString::copy.
• Add CHAR statistics serialization using
  AlignedRowWriter#writeChar.
• Add a serialization round-trip test covering min/max values, null count,
  and a value shorter than the declared CHAR length.
• Add a predicate pruning test for a CHAR column.

Tests

• LogRecordBatchStatisticsCollectorTest#testCharStatisticsSerializationRoundTrip:
  verifies that CHAR(3) values, including the shorter value "a", are
  collected and serialized with the expected min/max values and null count.
• RecordBatchFilterTest#testCharStatisticsCanPruneRecordBatch:
  verifies that CHAR statistics can exclude a record batch that cannot
  satisfy the predicate.
• Targeted tests: 24 tests, 0 failures.
• mvn spotless:check -pl fluss-common,fluss-server passes.
• Checkstyle and Apache RAT validation pass.

API and Format

No public API or statistics format change.

The implementation uses the existing CHAR internal representation and
statistics serialization format.

Documentation

No documentation change; behavior-only bug fix.

• No generative AI tools used
• Yes (please specify the tool below)

Assisted by: OpenAI Codex following the guidelines.