Multiplication: add 2d matrix versions

[ghost]

Add:

LLSparseMatrix A with:

• [x] LLSparseMatrix B: A * B and A^t * B;
• [x] NumPy B: A * B and A^t * B;
• [ ] NumPy B: B * A and B * A^t;
• [ ] CSRSparseMatrix B: A * B and A^t * B;
• [ ] CSCSparseMatrix B: A * B and A^t * B;

What about CSRSparseMatrix? CSCSparseMatrix?

[dpo]

I don't think it is necessary to implement the remaining three. (For the second one, we have B*A = (A.T*B.T).T and B*A.T = (A*B.T).T)

[ghost]

After all the work I've put to use optimized code (well kind of), wouldn't it be a pity to rely on compositions of operations that are much slower?

I agree to put this on freeze for the moment and implement more important operations. If I have time, I'll come back on this.

[dpo]

Sorry I didn't realize that. Sure then, that's great!