Purpose

Linked issue: close #xxx

This change bundles two independent items:

1. Parquet byte-level read metrics (feature)

Add two counters so callers can observe how many bytes a Parquet read actually consumes:

• parquet.read.storage-read-bytes — physical bytes read from storage.

2. Revert #397 commit internal-retry / dedup (fix)

#397 changed the commit-exception branch in TryCommitOnce from return Status::Invalid(...) to
return false, driving an internal retry loop plus a CheckCommitted/retry_start_snapshot_id
dedup. This introduced flakiness in CleanInteTest.TestDropPartitionAndExpireSnapshotWithIOException
and diverged from Java, which has no equivalent RetryCommitResult type system in this C++ codebase.

This reverts to the pre-#397 behavior:

• Remove retry_start_snapshot_id from the TryCommit loop and the TryCommitOnce signature.
• Remove the CheckCommitted method (definition and declaration).
• Restore the commit-exception branch to
  return Status::Invalid("You need call FilterAndCommit to retry commit for exception. ", ...).

Rationale: on a commit exception the outcome is uncertain (an atomic write may have timed out but
actually landed). Retrying internally can produce a duplicate snapshot. Safe retry belongs to the
upper layer via FilterAndCommit, which dedups by CommitIdentifier(); a bare Commit retry has no
such dedup and would create a duplicate. Returning Status::Invalid makes that contract explicit and
keeps guard.Release() so freshly written meta files are not cleaned up.

Tests

• TestCommitWithAtomicWriteSnapshotTimeoutAndActuallySucceed restored to its pre-feat(commit): align functionality with java paimon commit #397 assertions:
  the initial Commit returns an error (ASSERT_NOK) while the snapshot has physically landed, and
  recovery goes through FilterAndCommit, which idempotently filters the already-committed
  identifier (num_committed == 0).
• CleanInteTest.TestDropPartitionAndExpireSnapshotWithIOException — re-enabled (removed the
  DISABLED_ prefix); the CI-friendly randomized loop step is kept. Verified deterministically by
  running the loop exhaustively (i += 1, 0..1000) locally: PASSED (~14 min), confirming the revert
  root-causes and eliminates the flakiness rather than lowering its probability.
• Parquet metric wiring covered by updated unit tests: parquet_file_batch_reader_test.cpp,
  page_filtered_row_group_reader_test.cpp, predicate_pushdown_test.cpp (Create call sites now pass
  the storage_read_bytes counter).

API and Format

• No public API change: all touched declarations are internal (src/paimon/...); no headers under
  include/paimon/ are modified.
• No storage format or protocol change.
• New observable metric counter key: parquet.read.storage-read-bytes.
• Internal signature changes: ParquetFileBatchReader::Create/constructor gain a required
  storage_read_bytes parameter; FileStoreCommitImpl::TryCommitOnce drops retry_start_snapshot_id
  and CheckCommitted is removed. All call sites are updated in this change.

Documentation

No new user-facing feature documentation. The change adds observability counters and reverts an
internal retry behavior; no doc pages are introduced.

Generative AI tooling