[WIP] Support low shuffle merge with multithreaded Parquet reader

[liurenjie1024]

Fixes #11095.

Description

Low shuffle merge previously required the Parquet reader type to be PERFILE. Other reader
modes fell back to classic merge because the Delta 2.4 reader could not apply its per-file
deletion vectors in the multithreaded path.

This change enables low shuffle merge in every configured Parquet reader mode. It:

• removes the reader-mode gate from low shuffle merge planning;
• uses a Delta-aware multithreaded reader when a low-shuffle scan would otherwise select the
  coalescing reader, including explicit COALESCING and applicable AUTO scans;
• reuses cuDF's Parquet deletion-vector processing for both per-file and multithreaded reads,
  through a shared helper also used by newer Delta shims;
• keeps deletion vectors and row-group offsets aligned with each file/chunk and preserves cuDF's
  row-index output when synthetic Delta metadata columns are required;
• supports metadata-only scans where no physical Parquet columns are decoded; and
• keeps predicate pushdown disabled for low-shuffle target scans.

PERFILE continues to use the per-file reader. Low-shuffle scans configured for combined-file
reading use the multithreaded reader so deletion-vector state remains file-scoped. Each test in
delta_lake_low_shuffle_merge_test.py now explicitly parameterizes PERFILE, MULTITHREADED,
and COALESCING; the redundant focused combined-files test was removed.

Draft status:

• Run upstream CI.
• Address current early review feedback.

Validation:

• mvn -o -pl dist,integration_tests -am -DskipTests -DskipITs -Dbuildver=340 -Dscala2.12 package
• mvn -o -pl delta-lake/delta-24x -am -DskipTests -DskipITs -Dbuildver=340 -Dscala2.12 verify
• Spark 3.4.0 with Delta Lake 2.4.0: targeted standard-upsert reader matrix passed for all three
  reader settings (6 passed, 6 expected xfailed).
• Spark 3.4.0 with Delta Lake 2.4.0: metadata-only multithreaded delete matrix passed with all
  data columns used as partition columns (6 passed).

Checklists

Documentation

• Updated for new or modified user-facing features or behaviors
• No user-facing change

Testing

• Added or modified tests to cover new code paths
• Covered by existing tests
  (Please provide the names of the existing tests in the PR description.)
• Not required

Performance

• Tests ran and results are added in the PR description
• Issue filed with a link in the PR description
• Not required