Question about Reversible lithium plating: are small negative per-cycle values in “Plated [A.h]” expected? (LFP ~280 A·h)

[Danarbit]

When I enable reversible lithium plating in PyBaMM, my per-cycle summaries show
small negative values in Loss of capacity to negative lithium plating [A.h]
(typically around ±1e-7 A·h). The histogram is centered near zero and the
per-cycle time series oscillates around zero, while SEI capacity loss grows
monotonically.

Question: Are these small negative per-cycle values expected with the
reversible plating model (i.e., due to stripping exceeding plating over my
integration window), or do they indicate a post-processing/sign convention issue?