Fix scattercarpet calc

[rreusser]

This addresses @alexcjohnson's comment that scattercarpet was doing something weird and not handling array properties correctly. I mostly copied scattercarpet from scatterternary, but I wasn't quite able to reconstruct the removed line in this PR. I replaced it with arraysToCalcdata and it seems like things work fine though.

[etpinard]

That looks right. I should've caught that while reviewing that carpet PR. My mistake.

@rreusser would you mind a test case to 🔒 this down? Something similar to this suite and more precisely this case would more than suffice.

[rreusser]

I'm currently on my way to plotcon and accidentally checked my iPhone cable so I have no wifi today. Anyone is free to seize control of this PR if the fix is needed urgently, otherwise I'll lock this down asap.

[alexcjohnson]

Added a test (I want this for my gradient test cases). I did it through a full plot since I wanted to test restyle too. I actually thought it might not work on restyle, due to this line and the fact we didn't make module.arraysToCalcdata like we did in some modules... but, test passes (as does a sanity check in the dashboard) so I think we're OK, just may point to an optimization we can do later for style changes like this, perhaps whenever we get around to overhauling restyle/relayout.

[etpinard]

Oh that arraysToCalcdata check is a thing of the past 🔪

Restyle now does a recalc whenever an arrayOk attribute is or is set to an array.

[etpinard]

But yeah, it might be better to 🔪 that in a different PR.

💃