Make the much lighter ArrayInterfaceGPUArrays

[ChrisRackauckas]

Gives almost all GPU support in a pretty good way with almost no dependencies. I like it like that.
fix project tomls

[codecov]

Codecov Report

Merging #270 (699e129) into master (763554d) will not change coverage.
The diff coverage is n/a.

❗ Current head 699e129 differs from pull request most recent head 7eb246e. Consider uploading reports for the commit 7eb246e to get more accurate results

@@          Coverage Diff           @@
##           master    #270   +/-   ##
======================================
  Coverage    0.00%   0.00%           
======================================
  Files           9       9           
  Lines        1332    1328    -4     
======================================
+ Misses       1332    1328    -4     

Impacted Files	Coverage Δ
src/ArrayInterface.jl	0.00% <ø> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 763554d...7eb246e. Read the comment docs.

[ChrisRackauckas]

julia> @time_imports using ArrayInterfaceGPUArrays
     15.8 ms      ┌ Preferences
     23.9 ms    ┌ JLLWrappers
    168.3 ms  ┌ LLVMExtra_jll
      4.5 ms      ┌ CEnum
     97.3 ms    ┌ LLVM
      1.7 ms    ┌ Adapt
    803.2 ms  ┌ GPUArrays
      2.8 ms  ┌ ArrayInterfaceCore
   1004.3 ms  ArrayInterfaceGPUArrays

compared to

#211 (comment)

julia> @time_imports using ArrayInterfaceCUDA
      2.6 ms  ┌ Compat
      1.5 ms  ┌ Requires
     52.3 ms    ┌ ChainRulesCore
     60.7 ms  ┌ AbstractFFTs
     15.7 ms      ┌ Preferences
     17.6 ms    ┌ JLLWrappers
    173.7 ms  ┌ LLVMExtra_jll
      1.2 ms  ┌ Reexport
     23.5 ms    ┌ RandomNumbers
     38.3 ms  ┌ Random123
      4.0 ms      ┌ CEnum
    100.4 ms    ┌ LLVM
      2.0 ms    ┌ Adapt
    799.5 ms  ┌ GPUArrays
      1.3 ms  ┌ ExprTools
      1.1 ms  ┌ IfElse
     34.2 ms    ┌ Static
      2.3 ms    ┌ ArrayInterfaceCore
    381.2 ms  ┌ ArrayInterface
     24.7 ms  ┌ TimerOutputs
      1.7 ms  ┌ CompilerSupportLibraries_jll
      6.2 ms  ┌ BFloat16s
    193.5 ms  ┌ GPUCompiler
   1059.9 ms  ┌ CUDA
   2855.1 ms  ArrayInterfaceCUDA