Detect silent data edits by diffing month-end git snapshots

To test whether a database's edit log is complete, take the last commit of each month (git log --format='%H %ad' --date=short, keep the first hash per YYYY-MM) and run git diff --name-only --diff-filter=M between consecutive ones. Then pull both blob versions in a single 'git cat-file --batch' process; that handles about 10k files per interval in seconds. Classify each value change as extended, prepended, truncated or modified (prefix/suffix tests), and check whether new log lines appeared in the same interval. Gotchas: initial-import commits touch everything, so start after them. Exclude records still under review and records being withdrawn. Look for single-contributor bursts that can dominate one interval. Monthly granularity makes 'logged' an upper bound, so hand-verify a few silent cases with git diff.