Fix method ambiguities

[devmotion]

Fixes #481.

I assume the only questionable change is that fixing the method ambiguity of map(t::Transform, ::AbstractVector), caused by map(::Tf, ::SparseVector) where {Tf} in SparseArrays, requires adding a dependency on SparseArrays.

[codecov]

Codecov Report

Patch coverage: 80.00% and project coverage change: -0.10 ⚠️

Comparison is base (9da7bfd) 94.25% compared to head (143c8b0) 94.16%.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #483      +/-   ##
==========================================
- Coverage   94.25%   94.16%   -0.10%     
==========================================
  Files          52       52              
  Lines        1375     1387      +12     
==========================================
+ Hits         1296     1306      +10     
- Misses         79       81       +2     

Impacted Files	Coverage Δ
src/chainrules.jl	87.65% <ø> (ø)
src/utils.jl	91.46% <50.00%> (ø)
src/transform/transform.jl	81.81% <75.00%> (-18.19%)	⬇️
src/kernels/overloads.jl	100.00% <100.00%> (ø)
src/transform/chaintransform.jl	81.48% <100.00%> (+1.48%)	⬆️

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

[devmotion]

I assume the only questionable change is that fixing the method ambiguity of map(t::Transform, ::AbstractVector), caused by map(::Tf, ::SparseVector) where {Tf} in SparseArrays, requires adding a dependency on SparseArrays.

I guess it's even more questionable since actually it is not sufficient to fix the method ambiguities in the tests which load additional packages (see, e.g., https://github.com/JuliaGaussianProcesses/KernelFunctions.jl/actions/runs/3268189406/jobs/5374299901#step:6:222 and https://github.com/JuliaGaussianProcesses/KernelFunctions.jl/actions/runs/3268189406/jobs/5374301250#step:6:224):

   Evaluated: isempty(Tuple{Method, Method}[(map(f::Tf, A::SparseArrays.AbstractCompressedVector) where Tf @ SparseArrays.HigherOrderFns /opt/hostedtoolcache/julia/nightly/x64/share/julia/stdlib/v1.9/SparseArrays/src/higherorderfns.jl:152, map(t::Transform, x::AbstractVector) @ KernelFunctions ~/work/KernelFunctions.jl/KernelFunctions.jl/src/transform/transform.jl:11), (map(f, A::NamedDims.NamedDimsArray) @ NamedDims ~/.julia/packages/NamedDims/kCxfn/src/functions.jl:157, map(t::Transform, x::AbstractVector) @ KernelFunctions ~/work/KernelFunctions.jl/KernelFunctions.jl/src/transform/transform.jl:11), (map(f, a1::StaticArraysCore.StaticArray, as::AbstractArray...) @ StaticArrays ~/.julia/packages/StaticArrays/PUoe1/src/mapreduce.jl:30, map(t::Transform, x::AbstractVector) @ KernelFunctions ~/work/KernelFunctions.jl/KernelFunctions.jl/src/transform/transform.jl:11), (map(f, As::AxisArray{T, N, D, Ax}...) where {T, N, D, Ax<:Tuple{Vararg{Axis}}} @ AxisArrays ~/.julia/packages/AxisArrays/v7vf4/src/core.jl:431, map(t::Transform, x::AbstractVector) @ KernelFunctions ~/work/KernelFunctions.jl/KernelFunctions.jl/src/transform/transform.jl:11), (map(f, A::AxisArray) @ AxisArrays ~/.julia/packages/AxisArrays/v7vf4/src/core.jl:422, map(t::Transform, x::AbstractVector) @ KernelFunctions ~/work/KernelFunctions.jl/KernelFunctions.jl/src/transform/transform.jl:11), (map(f::Tf, A::Union{SparseArrays.AbstractCompressedVector, SparseArrays.AbstractSparseMatrixCSC}, Bs::Vararg{Union{SparseArrays.AbstractCompressedVector, SparseArrays.AbstractSparseMatrixCSC}, N}) where {Tf, N} @ SparseArrays.HigherOrderFns /opt/hostedtoolcache/julia/nightly/x64/share/julia/stdlib/v1.9/SparseArrays/src/higherorderfns.jl:156, map(t::Transform, x::AbstractVector) @ KernelFunctions ~/work/KernelFunctions.jl/KernelFunctions.jl/src/transform/transform.jl:11)])

[devmotion]

So I am not sure if we can fix the map method ambiguity issues at all.

[theogf]

@devmotion Can we get this merged anyway (we can leave some ambiguities)

[theogf]

Lgtm when the tests are fixed

[devmotion]

@theogf I managed to fix all method ambiguities by completely removing the definition of map(t::Transform, x::AbstractVector) (causes problems with basically every array package since they tend to implement map(f, x::MyArray)). Instead I only defined map(t::Transform, x::Union{ColVecs,RowVecs}).