fix linbox solve for sparse matrices

[videlec]

All the issues had been treated in #23214 which get rids of the sage-linbox interface.

---

While cleaning the linbox interface at #24544 we encountered some issues with an innocent type change. The sage-linbox interface is using LinBox::DenseVector<Givaro::Modular<unsigned int>> while it would be more natural to deal with Givaro::Modular<int64_t> or Givaro::Modular<uint64_t>. Though some strange compilation failures were present on macOS X, see discussion from the discussion at #24544.

Once this is fixed, we could get rid of the sage-linbox interface.

This problem is also tracked in linbox at issue 117.

Upstream: Reported upstream. No feedback yet.

CC: @ClementPernet @fchapoton

Component: interfaces

Issue created by migration from https://trac.sagemath.org/ticket/26178

[videlec]

Upstream: Reported upstream. No feedback yet.

[fchapoton]

comment:6

setting this to positive

[embray]

comment:7

Presuming these are all correctly reviewed as either duplicate, invalid, or wontfix.