Introduce Julia type to encapsulate spmat

[dpo]

The QRMumps function calls tend to be long, Fortran-like and difficult to decipher. I think a Julia type to encapsulate the spmat type would greatly enhance readability and user friendliness. For example, if A is an spmat, it would be good to be able to write

mul!(y, A, x)   # = qrm_spmat_mv!(A, 1, y, 0, x, transp = 'n')
mul!(y, A’, x)  # = qrm_spmat_mv!(A, 1, y, 0, x, transp = 't')

and have A * x and A’ * x work similarly after allocating new storage.

In the same vein, we could implement ldiv!(y, A, b), A \ b, A’ \ b, etc.

What do you think @MaxenceGollier, @amontoison and @abuttari ?

[amontoison]

It's a good idea Dominique. I defined a few Julia functions like \ or ldiv! but only for the factorization (spfct).