[SPARK-56462][SQL]Fix UPDATE/INSERT schema evolution when source col contains dot

[jiwen624]

What changes were proposed in this pull request?

In Analyzer.scala, the UPDATE * and INSERT * schema-evolution paths inside ResolveReferences built assignment keys for new source columns using UnresolvedAttribute(sourceAttr.name). That constructor splits the name on dots, so a column named `job.title` produced nameParts = Seq("job", "title") instead of Seq("job.title").

The fix replaces both call sites with UnresolvedAttribute.quoted(sourceAttr.name), which treats the whole name as a single quoted identifier.

Why are the changes needed?

With a dot in the source column name, isSameColumnAssignment compared a two-part key against a single-part value and returned false, blocking schema evolution. Analysis then failed with UNRESOLVED_COLUMN.

Does this PR introduce any user-facing change?

Yes. MERGE INTO … WHEN MATCHED UPDATE * WHEN NOT MATCHED INSERT * with schema evolution now correctly handles source columns whose names contain a dot.

How was this patch tested?

Added a testEvolution case in MergeIntoSchemaEvolutionBasicTests covering both schema-evolution-enabled (dot-named column is added and populated) and schema-evolution-disabled (extra column is silently ignored).

Was this patch authored or co-authored using generative AI tooling?

Yes.

[jiwen624]

@dongjoon-hyun @cloud-fan @HyukjinKwon could you let me know your thoughts on this fix? Thanks