Latexify sparse arrays

[gustaphe]

Close #269

Sparse arrays did not work. Since that's a stdlib, I figured it was uncontroversial to using it.
This is significant because it's our first "internal recipe". There's probably several types we currently have hard-coded that could be made into recipes like this.

Also, in the future someone might do something clever about sparse matrices, like only printing significant zeroes etc. This just turns it into a normal matrix before latexifying.

[ChrisRackauckas]

Since that's a stdlib

It's not anymore 😅 , and it is controversial because it has a GPL dependency. I am recommending all libraries make SparseArrays into an extension, so I'll also recommend that here.

[gustaphe]

Ach, I was hoping to avoid that, but I guess it's cleaner anyway. Thanks for the heads up.

[gustaphe]

Since that's a stdlib

It's not anymore 😅

Someone should tell the docs

[ChrisRackauckas]

Maybe it's still there, but it's:

1. No longer in the system image, so it's not loaded by default
2. It's maintained in a separate repository in a different Github organization
3. Importantly, Suitesparse is not loaded until it's loaded (which wasn't true until v1.10)
4. You can soon (v1.12?) build system images without it in there (making a GPL free system image)

For backwards compatibility reasons I think it'll have to continue coming with the standard Julia installation, but it effectively is just a separate package that is now shipped with the standard binary and not even in the image.

[gustaphe]

It functionally doesn't matter, I was just rationalizing my own laziness. Always interesting to learn a bit more about the nuts and bolts though.