Baseline status regressions should require a note

[ddbeck]

Blocked on #915.

If we have a Baseline status regression (see #1947), we ought to be transparent with data consumers and developers about why those regressions happened.

Such notes probably ought to have, at a minimum:

• Some sort type information (e.g. "status-regression"), so tooling can distinguish them from other kinds of notes.
• The date of the status change (e.g., "2024-10-15"), so tooling needn't detect (and authors needn't figure out how to present) when a feature changed.
• An explanation of why the status changed (e.g, "A serious bug in [browser] caused this feature to be downgraded from highly available to newly available.", "A data error caused this feature to be wrongly reported as Baseline.", etc.)
• A reference URL (to a browser bug, a compat data change in BCD, etc.)