fix: make (multi)basis work on CuArray

[gdalle]

Combining this small fix with gdalle/SparseMatrixColorings.jl#253, we can perform sparse autodiff on NVIDIA GPUs and leverage CUDA's native sparse matrices:

using ADTypes
using CUDA, CUDA.CUSPARSE
using DifferentiationInterface
using ForwardDiff: ForwardDiff
using LinearAlgebra
using SparseMatrixColorings

x = CuVector(rand(Float32, 10))
S = CuSparseMatrixCSR(Diagonal(ones(Float32, 10)))

sparsity_detector = ADTypes.KnownJacobianSparsityDetector(S)  # make sure the sparsity pattern is a CUDA sparse matrix
coloring_algorithm = GreedyColoringAlgorithm()
backend = AutoSparse(AutoForwardDiff(); sparsity_detector, coloring_algorithm)

julia> prep = prepare_jacobian(identity, backend, x);

julia> jacobian(identity, prep, backend, x)
10×10 CuSparseMatrixCSR{Float32, Int32} with 10 stored entries:
 1.0   ⋅    ⋅    ⋅    ⋅    ⋅    ⋅    ⋅    ⋅    ⋅ 
  ⋅   1.0   ⋅    ⋅    ⋅    ⋅    ⋅    ⋅    ⋅    ⋅ 
  ⋅    ⋅   1.0   ⋅    ⋅    ⋅    ⋅    ⋅    ⋅    ⋅ 
  ⋅    ⋅    ⋅   1.0   ⋅    ⋅    ⋅    ⋅    ⋅    ⋅ 
  ⋅    ⋅    ⋅    ⋅   1.0   ⋅    ⋅    ⋅    ⋅    ⋅ 
  ⋅    ⋅    ⋅    ⋅    ⋅   1.0   ⋅    ⋅    ⋅    ⋅ 
  ⋅    ⋅    ⋅    ⋅    ⋅    ⋅   1.0   ⋅    ⋅    ⋅ 
  ⋅    ⋅    ⋅    ⋅    ⋅    ⋅    ⋅   1.0   ⋅    ⋅ 
  ⋅    ⋅    ⋅    ⋅    ⋅    ⋅    ⋅    ⋅   1.0   ⋅ 
  ⋅    ⋅    ⋅    ⋅    ⋅    ⋅    ⋅    ⋅    ⋅   1.0

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 97.93%. Comparing base (d8905f5) to head (34019b5).
Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #810      +/-   ##
==========================================
- Coverage   97.93%   97.93%   -0.01%     
==========================================
  Files         128      128              
  Lines        7693     7690       -3     
==========================================
- Hits         7534     7531       -3     
  Misses        159      159              

Flag	Coverage Δ
DI	98.81% <100.00%> (-0.01%)	⬇️
DIT	95.61% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.