Another lnlq NaN case

```julia julia> A = [-1.0 0] 1×2 Matrix{Float64}: -1.0 0.0 julia> b = [-3.0] 1-element Vector{Float64}: -3.0 julia> lnlq(A, b) ([NaN, NaN], [NaN], LNLQStats niter: 2 solved: true residuals: [] error with bnd: false error bnd x: [] error bnd y: [] timer: 17.07μs status: solutions (xᶜ, yᶜ) good enough for the tolerances given ) julia> cgne(A, b) ([3.0, 0.0], SimpleStats niter: 1 solved: true inconsistent: false indefinite: false npcCount: 0 residuals: [] Aresiduals: [] κ₂(A): [] timer: 16.69μs status: solution good enough given atol and rtol ) ```