`warpx.use_filter` produces strange results in RZ FDTD mode

[RemiLehe]

It seems that using warpx.use_filter=1 produces strange results, for the RZ FDTD solver. See for instance the snapshot for a laser-wakefield simulation below: (note: labels are swapped; this shows the charge density rho)

Maybe we should add an assert to prevent users from selecting this option.

Note by @EZoni: The CI test LaserAccelerationRZ has warpx.use_filter = 1 in this configuration (RZ FDTD). This would need to be modified, with subsequent benchmark reset, if we add an assert as above.

[EZoni]

Is there an input file that we can use to reproduce this issue?

[dpgrote]

In PR #2031, filtering is turned off by default with RZ FDTD. If filtering is turned on with RZ FDTD, a warning will be printed since it has not been implemented. Can this issue be closed or left open as a reminder that filtering in not implemented in RZ FDTD?

[ax3l]

Until we can look into this properly, #2775 will add an abort for now.

[roelof-groenewald]

When adding PEC boundary handling to Cartesian geometry cases (see above link), strange behavior was observed when applying the boundary correction before filtering and therefore the boundary handling had to be done after filtering. The RZ volume and boundary correction is applied before filtering. Should that also be moved to happen afterwards instead?