Account for Carried-Over Open Findings in pr_review.py's Shortfall Comparison #2481

Problem

pr_review.py compares a Copilot round's stated finding total against the threads that round opened (round_threads), and prints a shortfall where the total is larger. The round_threads docstring records the open doubt: if the overview's open count carries findings an earlier round raised and left open, the comparison prints a standing shortfall about findings that already have threads. It names what would settle it as a round whose open count exceeds the threads it opened.

Such rounds exist. Measured over recent ccr-overview-v2 rounds on this repository and one sibling, most state an open count equal to the threads the round opened, and a few do not:

• ProjectTemplate Align details Tag Boundaries Across pr_review.py's Readers #2415: a later round states 2 open and opened 0 threads.
• ProjectTemplate State the Archive Registration in host-setup.md's Fleet Skills Install #2391: a later round states 1 open and opened 0 threads.

The format's 2026-10-07 revision states the total as a bare **N open findings** line beside a N resolved since last review section, which reads as a running count of open findings across rounds rather than what one round raised.

Done when

• The shortfall comparison accounts for open findings an earlier round raised, or is scoped so a carried-over finding with its own thread does not print as one with no thread.
• A fixture of a later round restating an earlier round's open finding covers it, with constructed titles and ids.