Add constructor for GPUArrays

[ldeso]

Currently constructing ComponentArrays on the GPU with for example ComponentArray(a=CUDA.ones(2,3), b=CUDA.ones(4,5)) fails with an error.

ERROR: CuArray only supports element types that are stored inline

This fails because constructing the ComponentArray causes the creation of an array of arrays, which is not possible on the GPU. This happens in the pushcat! call at line 179 in componentarray.jl:

ComponentArrays.jl/src/componentarray.jl

Lines 178 to 182 in a7d6f7d

	function make_idx(data, x::AbstractArray, last_val)
	pushcat!(data, x)
	out = last_index(last_val) .+ (1:length(x))
	return (data, ViewAxis(out, ShapedAxis(size(x))))
	end

This pull request fixes the issue by constructing the ComponentArray on the CPU and moving it back to the GPU afterwards.

Fixes #158. Needed for SciML/NeuralPDE.jl#594.

[codecov-commenter]

Codecov Report

Merging #166 (61801da) into master (a7d6f7d) will decrease coverage by 0.62%.
The diff coverage is 0.00%.

@@            Coverage Diff             @@
##           master     #166      +/-   ##
==========================================
- Coverage   77.13%   76.50%   -0.63%     
==========================================
  Files          20       20              
  Lines         608      613       +5     
==========================================
  Hits          469      469              
- Misses        139      144       +5     

Impacted Files	Coverage Δ
src/compat/gpuarrays.jl	65.78% <0.00%> (-9.97%)	⬇️

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[jonniedie]

Cool, thanks!