[SPARK-57064][SQL] Widen bucketing rule pattern matches to use FileSourceScanLike trait

[rdtr]

Description:

Summary

• Widen 3 read-only pattern match sites in DisableUnnecessaryBucketedScan and CoalesceBucketsInJoin from concrete FileSourceScanExec to the FileSourceScanLike trait
• Two match sites that call .copy() are intentionally left on FileSourceScanExec
• Enables third-party columnar plugins (Gluten, Comet, RAPIDS) to be recognized by bucketing optimization rules

Test plan

• Existing bucketing test suites pass (no behavioral change for vanilla Spark since FileSourceScanExec already extends FileSourceScanLike)
• Verified all accessed fields (bucketedScan, relation, optionalNumCoalescedBuckets) are declared on the FileSourceScanLike trait