as titled avoid parallel goroutine in MergeReaders
bonus: update golang ver

Changelog

• Change: Enforced strict, deterministic file processing order for file→DB runs.

  • --src-stream file://<folder> and --src-stream file://<folder>/* are now processed sequentially in ascending filename order.
  • Replaces prior async merging via goroutines that could interleave rows and produce non-deterministic ordering.

• Impact

  • Ordering rule: lexical by filename after discovery and sort. Use zero-padded names (e.g., tablename_YYYYMMDD.csv, tablename_0001.csv) to match chronological order.
  • Scope: applies to Proton and all DB targets; also affects --stdout.
  • Performance: slight throughput trade-off vs previous async merging in exchange for correctness.

• Verify

  1. Generate 50 files: tablename_20240101.csv … tablename_20240150.csv (each with header and several rows).
  2. On the old build, run with --src-stream file://<folder>/* targeting Proton.
     • Expected: rows written to Proton may be out of order/non-deterministic.
  3. On the new build, run the same.
     • Expected: rows are written in strict ascending filename order (1 → 50).