R4. Neither a reference nor a repair transfers

The ramp handoff's single scalar vertex does not supply a second coefficient with a proved charge $r_{\epsilon }$ Ramp. The two-qubit program in Option B still waits for an independent reference and, for whole-algebra invariance, the matching condition (eq:matching). It is not retargeted at ${\mathrm{Obs}}_{f,g,\epsilon }^{\mathrm{R}}$.

The ramp source likewise excludes the sharp-regulator limit. Its smooth ramp kernel is not the convolution defect. No Enter operation changes either

A change to either coordinate prescription requires a new classical covariance argument, not a transfer of a readout theorem.