bench(csharp-query): report scan source registrations

[s243a]

Summary

• Adds compact report-tsv and markdown output modes to benchmark_scan_materialization.py.
• Summarizes scan benchmark source-mode results by scale/mode, including medians, row counts, source registrations,
  and artifact bucket strategies.
• Keeps the existing detailed TSV output as the default.
• Documents using the report formats for join,nary_join source-registration comparisons.

Verification

• python3 -m py_compile examples/benchmark/benchmark_scan_materialization.py
• git diff --check
• python3 examples/benchmark/benchmark_scan_materialization.py --scales 300 --modes join,nary_join --source-modes auto,preload,artifact-prebuilt --strategies auto --repetitions 1 --format markdown
• python3 examples/benchmark/benchmark_scan_materialization.py --scales 300 --modes nary_join --source-modes preload --strategies auto --repetitions 1