update values before klu_factor

Otherwise it's just using the values from before!

BTW, why do we need to klu_factor before klu!, instead of just klu!? Don't that do the factorization as well?

Author: ChrisRackauckas

src/factorization.jl

@@ -143,6 +143,7 @@ function SciMLBase.solve(cache::LinearCache, alg::KLUFactorization; kwargs...)
             # If we have a cacheval already, run umfpack_symbolic to ensure the symbolic factorization exists
             # This won't recompute if it does.
             KLU.klu_analyze!(cache.cacheval)
+            copyto!(cache.cacheval.nzval,A.nzval)
             if cache.cacheval._numeric === C_NULL # We MUST have a numeric factorization for reuse, unlike UMFPACK.
                 KLU.klu_factor!(cache.cacheval)
             end