Fixes for staggering in lapd-drift example

[johnomotani]

Also replaces deprecated Grad_par_CtoL() and Grad_par_LtoC() operators.

[ZedThree]

@ZedThree, @bendudson The staggered derivative operators in lapd_drift.cxx were being used the opposite way round to what I thought they meant. Grad_par_CtoL() was taking a flux (which I expected to be at CELL_YLOW) as argument, and returning a result for a scalar equation (which I expected to be at CELL_CENTRE), I thought it should be the other way around: i.e. CtoL meant 'argument at C' and 'result at L'. Doesn't really matter now that the CtoL and LtoC operators are deprecated anyway, but noting here in case issues come up converting old physics models.

Originally posted by @johnomotani in #1923 (comment)

I double checked the old implementation of the LtoC/CtoL operators, and they did work as you thought. A little bit worrying! Checking a couple of other old physics models, and they are mostly using them the correct way round. I think only examples/tokamak-2fluid/2fluid.cxx and examples/elm-pb/elm-pb.cxx are also backwards.

@johnomotani Please could you check those models too?

EDIT: most of the examples that use the LtoC/CtoL operators don't actually call setLocation, so they're probably all a little bit dubious

[johnomotani]

@johnomotani Please could you check those models too?

EDIT: most of the examples that use the LtoC/CtoL operators don't actually call setLocation, so they're probably all a little bit dubious

I guess originally as long as the model was consistent with itself it didn't matter much. Having LtoC/CtoL 'back to front' would just correspond to having the scalars at CELL_YLOW and fluxes at CELL_CENTRE. I think now it would make sense to update all of them to the current conventions, ignoring which way round CtoL/LtoC were originally. I'll make a PR.