[WIP] Fix bug in energy-preserving explicit electrostatic PIC-MCC algorithm

[EZoni]

This may fix a bug that @kale-j observed in an electromagnetic simulation based on #6275, and that was troubleshooted by @oshapoval, @RemiLehe, and myself.

This PR takes care of the bug fix for the electrostatic case, originally implemented in #5955. The bug fix for the electromagnetic case is included in #6275.

This change may break a number of CI tests, so I flag the PR as "[WIP]" until the CI checks pass.

[EZoni]

@RemiLehe @oshapoval

Some CI tests failed here, because of the particle boundary change:

• 1d:

The following tests FAILED:
	 26 - test_1d_ionization_neutral_dsmc.analysis (Failed)
	 27 - test_1d_ionization_neutral_dsmc.checksum (Failed)
	 30 - test_1d_charge_exchange_dsmc.analysis (Failed)
	 31 - test_1d_charge_exchange_dsmc.checksum (Failed)
	 35 - test_1d_photoneutralization_dsmc.checksum (Failed)
	 39 - test_1d_two_product_reaction_dsmc.checksum (Failed)

• cartesian_3d:

The following tests FAILED:
	222 - test_3d_linear_compton_bunch_laser.checksum (Failed)

I think this list is reasonably similar to the list of tests that used to fail in #5955.

[EZoni]

Following up on #6453 (comment), this is the analysis plot for test_3d_linear_compton_bunch_laser:

@aeriforme Does it look okay to you? Note that the test analysis passes, only the checksums fail.

[aeriforme]

Following up on #6453 (comment), this is the analysis plot for test_3d_linear_compton_bunch_laser:

@aeriforme Does it look okay to you? Note that the test analysis passes, only the checksums fail.

It looks good. But why did the checksums change so much? The maximum relative error is 4.74e+00.
Could you explain what bug this PR fixes?

[EZoni]

@aeriforme

The issue this mitigates was observed by @kale-j in an electromagnetic simulation based on #6275 and looked like plasma building up anomalously at the simulation boundary.

@oshapoval, @RemiLehe and I believe the change should be implemented for both the electromagnetic and electrostatic case, hence this PR. However, we may be wrong.