Add a generic scal!

[amontoison]

scal! was only defined for Float32, Float64, ComplexF32 and ComplexF64 vectors.

[N5N3]

I'm not sure it a good idea to bring more BLAS api into LinearAlgebra.
Generic scal(!)(a, X) could be easily replaced with a * X and mul!(X, a, X).
The only problem needs to concern is #43153 (mul!(X, a, X) is a self-inplace broadcast).
But it could be fixed easily after we have purity inferability.

[ViralBShah]

I tend to agree with @N5N3's comment. Closing for now, but happy to reopen if there's a good reason.

[amontoison]

@N5N3 @ViralBShah
Do we have a generic function to do a * X in-place now?
My main concern is how to add a dispatch on GPU arrays and I would like to avoid broadcast.

[ViralBShah]

I thought we have a mul! method for this case.

julia> a
1

julia> X = rand(10,10);

julia> Y = zeros(10,10);

julia> @which mul!(Y,a,X)
mul!(C, A, B)
     @ LinearAlgebra ~/.julia/juliaup/julia-1.11.0+0.aarch64.apple.darwin14/share/julia/stdlib/v1.11/LinearAlgebra/src/matmul.jl:253