Added multiplication of transpose / adjoint matrices by diagonal matrices

[amontoison]

Updated version of #2482

[maleadt]

CI failures look related.

[amontoison]

Julia wants to dispatch to another method with Transpose and Adjoint:

mul!(d_YA, d_Y, transpose(d_A))

If you want to allow scalar iteration, use `allowscalar` or `@allowscalar`
to enable scalar iteration globally or for the operations in question.
Stacktrace:
  [1] error(s::String)
    @ Base ./error.jl:35
  [2] errorscalar(op::String)
    @ GPUArraysCore ~/.julia/packages/GPUArraysCore/GMsgk/src/GPUArraysCore.jl:155
  [3] _assertscalar(op::String, behavior::GPUArraysCore.ScalarIndexing)
    @ GPUArraysCore ~/.julia/packages/GPUArraysCore/GMsgk/src/GPUArraysCore.jl:128
  [4] assertscalar(op::String)
    @ GPUArraysCore ~/.julia/packages/GPUArraysCore/GMsgk/src/GPUArraysCore.jl:116
  [5] getindex
    @ ~/.julia/packages/GPUArrays/qt4ax/src/host/indexing.jl:50 [inlined]
  [6] macro expansion
    @ ~/Applications/julia/julia-1.10.5/share/julia/stdlib/v1.10/LinearAlgebra/src/diagonal.jl:318 [inlined]
  [7] macro expansion
    @ ./simdloop.jl:77 [inlined]
  [8] __muldiag!(out::CuArray{Float64, 2, CUDA.DeviceMemory}, D::Diagonal{ComplexF64, CuArray{…}}, B::Transpose{ComplexF64, CuArray{…}}, _add::LinearAlgebra.MulAddMul{true, true, Bool, Bool})
    @ LinearAlgebra ~/Applications/julia/julia-1.10.5/share/julia/stdlib/v1.10/LinearAlgebra/src/diagonal.jl:317
  [9] _mul_diag!
    @ ~/Applications/julia/julia-1.10.5/share/julia/stdlib/v1.10/LinearAlgebra/src/diagonal.jl:391 [inlined]
 [10] _mul!
    @ ~/Applications/julia/julia-1.10.5/share/julia/stdlib/v1.10/LinearAlgebra/src/diagonal.jl:397 [inlined]
 [11] mul!
    @ ~/Applications/julia/julia-1.10.5/share/julia/stdlib/v1.10/LinearAlgebra/src/bidiag.jl:427 [inlined]
 [12] mul!(C::CuArray{Float64, 2, CUDA.DeviceMemory}, A::Diagonal{ComplexF64, CuArray{ComplexF64, 1, CUDA.DeviceMemory}}, B::Transpose{ComplexF64, CuArray{ComplexF64, 2, CUDA.DeviceMemory}})
    @ LinearAlgebra ~/Applications/julia/julia-1.10.5/share/julia/stdlib/v1.10/LinearAlgebra/src/matmul.jl:237
 [13] top-level scope
    @ REPL[71]:1
Some type information was truncated. Use `show(err)` to see complete types.

@dkarrasch Do you have an idea how I can fix it?
Should I add a method for generic_matmatmul! instead?

[dkarrasch]

Tests seem to pass? Or where is the problem? I think you're overloading the right method, which is the three-arg mul! in line [12]. That one is completely generic, so any method that specifies the arguments (to arrays that it owns) should have precedence.

[amontoison]

I forgot to change the branch when I tested the PR on a cluster...
The issue was between the chair and the screen (me) 😄
Thanks for confirming that I overloaded the right method.