Add KnownJacobianSparsityDetector and KnownHessianSparsityDetector

[adrhill]

Checklist

• Appropriate tests were added
• Any code changes were done in a way that does not break public API
• All documentation related to code changes were updated
• The new code follows the
  contributor guidelines, in particular the SciML Style Guide and
  COLPRAC.
• Any new documentation only uses public API

Additional context

Adds KnownJacobianSparsityDetector and KnownHessianSparsityDetector, closing #80.

[gdalle]

This is exactly how I would have coded it too, thanks for this contribution!

[adrhill]

Are we ok with the KnownJacobian/KnownHessian prefix?
Another option that comes to my mind is ManualJacobian. SparseDiffTools calls it JacPrototype but I find the others to be more informative.

[gdalle]

I would prefer Fixed to Known

[gdalle]

Let's wait until @ChrisRackauckas gives an opinion on this

[adrhill]

I would prefer Fixed to Known

Then let me throw Constant into the ring as well. ;)

[ChrisRackauckas]

I'd go with Known. Constant or Fixed isn't really correct, because that's more information about the sparsity pattern itself rather than information about whether it's already known. I could see constant or fixed information being used as a specialization to know that you don't need to handle certain elements of conditional handling in the sparsity detection (i.e. the concolic execution of the original algorithm) if you knew that you always had a fixed pattern and don't need to take the union.

[gdalle]

Let's go with Known then.