fix(irf): stats command undercounts newly added tail items

[4444J99]

Failure state

The source file INST-INDEX-RERUM-FACIENDARUM.md contains a newly added open item:

• IRF-SYS-044

But organvm irf stats still reports the old totals immediately afterward:

• Total: 715
• Open: 384
• SYS: 29

Expected from file contents after insertion:

• Total: 716
• Open: 385
• SYS: 30

Why this matters

The IRF CLI is being used as a governance truth source during session closeout. If it silently undercounts new items, operators can incorrectly conclude the registry is synchronized when it is not.

Reproduction

1. Add a new IRF-SYS-* row near the tail of INST-INDEX-RERUM-FACIENDARUM.md.
2. Run organvm irf stats.
3. Compare CLI totals to the file contents.

Observed on

• Date: 2026-04-02
• Repo: meta-organvm/organvm-corpvs-testamentvm
• During closeout of public-record-data-scrapper reliability hardening session

Suggested investigation

• Check whether the IRF parser stops at a sentinel, duplicate header, or statistics block boundary.
• Add a parser test with a tail-appended item near the final open section.
• Fail loudly if the source file and derived counts disagree during refresh tooling.

[4444J99]

Backlog engagement 2026-07-19 — disposition: distilled.

Verified delivered on current main: src/organvm_engine/irf/parser.py supports compact hand-appended tail rows, cmd_irf_stats reports parse-incomplete diagnostics instead of silently trusting undercounts, and tests/test_irf_cli.py::test_irf_stats_counts_tail_row_without_leading_pipe covers the tail-row stats case.