Drift-detection seam. The DEFAULT implementation (RowCountProxyDriftDetector)
is intentionally simple and HONEST — it uses an observable proxy (relative
change in scored-row volume vs the training row count) rather than pretending to
do a real distributional test. A genuine statistical drift method
(population-stability index / KS over the score distribution) is a flagged
follow-up (plan §16 [O]); inject a real detector here when it lands.
Drift-detection seam. The DEFAULT implementation (RowCountProxyDriftDetector) is intentionally simple and HONEST — it uses an observable proxy (relative change in scored-row volume vs the training row count) rather than pretending to do a real distributional test. A genuine statistical drift method (population-stability index / KS over the score distribution) is a flagged follow-up (plan §16 [O]); inject a real detector here when it lands.