The implementation of the particle CFL option has been updated for the implicit PIC:
#6428

The same fix is more difficult for the explicit PIC, because the collision happens in the middle of a step. As a consequence, the following problematic scenario could in principle happen:

• At the beginning of the PIC loop, $\Delta t$ is determined based on the current max velocity of the particles
• The first half-push of position happens without issues
• The collision happens. This could in principle result in some particles getting much larger velocity, such that, with the current $\Delta t$, the particles would go out of the guard cells in a single half-timestep.
• The second half-push happens and some particles get outside of the guard cells.
• Charge deposition happens, and could result in out-of-bound.