Relax type annotations in sparse prep

[ErikQQY]

Part of #646

Relax type annotations of the compressed matrix from AbstractMatrix{<:Real} to AbstractMatrix{<:Number} in PushforwardSparseJacobianPrep, PullbackSparseJacobianPrep, SparseHessianPrep and MixedModeSparseJacobianPrep to support sparse Jacobian/Hessian preparation for complex numbers.

The test case for this PR is adapted from MWE, so need some advice on this.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 97.52%. Comparing base (fd3dc14) to head (c5e8561).
Report is 2 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #647   +/-   ##
=======================================
  Coverage   97.52%   97.52%           
=======================================
  Files         111      111           
  Lines        5536     5536           
=======================================
  Hits         5399     5399           
  Misses        137      137           

Flag	Coverage Δ
DI	98.65% <ø> (ø)
DIT	95.35% <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.

[gdalle]

Thanks! I can take over for the tests

[gdalle]

I assume you were using complex numbers with FiniteDiff? I tried with ForwardDiff but it won't even build the Duals

[ErikQQY]

Yeah, it should be FiniteDiff, very sorry I may not have specified this clearly. When handling complex numbers, in my specific use case, according to the input x type(ComplexF64, etc.) in prepare_jacobian, FiniteDiff backend should be used. But the general prepare_jacobian API failed to handle this no matter which backend is being used.

[ErikQQY]

Can we cut a release for this patch?

[gdalle]

Here you go: JuliaRegistries/General#120422