Change FEMSWEEP to read in file for mesh data

[rchen20]

Summary

• This PR is a refactoring
• It does the following:
  • Refactors FEMSWEEP to read in a file to populate mesh connectivity data.

[MrBurmark]

Does reading the data also affect the problem size, number of elements, etc?

[artv3]

How big can meshes get? I'm wondering how the meshes will impact the repo size.

[MrBurmark]

This can probably be a free function, probably in an unnamed namespace.

[rchen20]

I can move it if some other kernel is going to use it, but I think the femsweep is the only one that requires something like this at the moment so I'll keep it the way it is.

[MrBurmark]

I just mean that it doesn't use any of the kernel's state, it could be a free function. Its not a big deal, but its probably best practice if that's true to make it a free function.

[rchen20]

@rhornung67 How about I revert this PR such that there is only one small mesh? That way the meshes won't show up in the git branches and history. After that, we can create another repo to store the larger meshes.

[artv3]

@rchen20 , you can also squash merge

[rchen20]

@rchen20 , you can also squash merge

Thanks, but I don't think squash merging will prevent people from downloading all the meshes on every git clone. I'd need to remove the meshes from the history entirely.

[artv3]

@rchen20 , you can also squash merge

Thanks, but I don't think squash merging will prevent people from downloading all the meshes on every git clone. I'd need to remove the meshes from the history entirely.

Ah I see, thanks for explaining. Sure I'm in favor of your original proposition