Change device! function parameter type from Int32 to Int

[simone-silvestri]

Following the discussion in JuliaGPU/KernelAbstractions.jl#636,
this PR aligns the type parameter of the device! function for CUDABackends to the one required in
https://github.com/JuliaGPU/KernelAbstractions.jl/blob/1ac546fc59cc611d749fa7a50e4a1efa3393851b/src/KernelAbstractions.jl#L622

[github-actions]

CUDA.jl Benchmarks

Benchmark suite	Current: 14a933e	Previous: f99784f	Ratio
latency/precompile	57559824690 ns	57505943660.5 ns	1.00
latency/ttfp	8221184268 ns	8175596902 ns	1.01
latency/import	4554132407 ns	4548330811 ns	1.00
integration/volumerhs	9627333.5 ns	9627247.5 ns	1.00
integration/byval/slices=1	147180 ns	147004 ns	1.00
integration/byval/slices=3	426079 ns	426015 ns	1.00
integration/byval/reference	145085 ns	144995 ns	1.00
integration/byval/slices=2	286565 ns	286391 ns	1.00
integration/cudadevrt	103657 ns	103484 ns	1.00
kernel/indexing	14124 ns	14144.5 ns	1.00
kernel/indexing_checked	14940 ns	14979 ns	1.00
kernel/occupancy	740.4360902255639 ns	696.1307189542483 ns	1.06
kernel/launch	2192.8888888888887 ns	2152.1111111111113 ns	1.02
kernel/rand	15803 ns	17139 ns	0.92
array/reverse/1d	19938 ns	20127 ns	0.99
array/reverse/2dL_inplace	66882 ns	66835 ns	1.00
array/reverse/1dL	70144 ns	70275 ns	1.00
array/reverse/2d	21951 ns	22347 ns	0.98
array/reverse/1d_inplace	11377.5 ns	9616 ns	1.18
array/reverse/2d_inplace	13269 ns	13357 ns	0.99
array/reverse/2dL	73882 ns	74201 ns	1.00
array/reverse/1dL_inplace	66741 ns	66737 ns	1.00
array/copy	21128 ns	21034 ns	1.00
array/iteration/findall/int	157964.5 ns	158193.5 ns	1.00
array/iteration/findall/bool	139673.5 ns	140393.5 ns	0.99
array/iteration/findfirst/int	161714 ns	161096 ns	1.00
array/iteration/findfirst/bool	162871 ns	161802 ns	1.01
array/iteration/scalar	73219 ns	72492 ns	1.01
array/iteration/logical	215922 ns	214800 ns	1.01
array/iteration/findmin/1d	51057.5 ns	50555 ns	1.01
array/iteration/findmin/2d	96635.5 ns	96485 ns	1.00
array/reductions/reduce/Int64/1d	43609 ns	43782 ns	1.00
array/reductions/reduce/Int64/dims=1	44757.5 ns	44868.5 ns	1.00
array/reductions/reduce/Int64/dims=2	61881 ns	61893.5 ns	1.00
array/reductions/reduce/Int64/dims=1L	89359 ns	89229.5 ns	1.00
array/reductions/reduce/Int64/dims=2L	88482 ns	88425 ns	1.00
array/reductions/reduce/Float32/1d	38068.5 ns	37869 ns	1.01
array/reductions/reduce/Float32/dims=1	52338 ns	52159 ns	1.00
array/reductions/reduce/Float32/dims=2	60427.5 ns	60004 ns	1.01
array/reductions/reduce/Float32/dims=1L	52804 ns	52689 ns	1.00
array/reductions/reduce/Float32/dims=2L	72590.5 ns	72514 ns	1.00
array/reductions/mapreduce/Int64/1d	43887 ns	44158 ns	0.99
array/reductions/mapreduce/Int64/dims=1	45179.5 ns	44917.5 ns	1.01
array/reductions/mapreduce/Int64/dims=2	62247 ns	61973 ns	1.00
array/reductions/mapreduce/Int64/dims=1L	89325 ns	89251 ns	1.00
array/reductions/mapreduce/Int64/dims=2L	88620 ns	88531 ns	1.00
array/reductions/mapreduce/Float32/1d	38044 ns	37472.5 ns	1.02
array/reductions/mapreduce/Float32/dims=1	51317 ns	52480 ns	0.98
array/reductions/mapreduce/Float32/dims=2	60731 ns	60249 ns	1.01
array/reductions/mapreduce/Float32/dims=1L	53299 ns	52856 ns	1.01
array/reductions/mapreduce/Float32/dims=2L	73266 ns	72524 ns	1.01
array/broadcast	20568.5 ns	20030 ns	1.03
array/copyto!/gpu_to_gpu	13272 ns	11386 ns	1.17
array/copyto!/cpu_to_gpu	213642 ns	216869 ns	0.99
array/copyto!/gpu_to_cpu	283807.5 ns	286424.5 ns	0.99
array/accumulate/Int64/1d	124845 ns	124830 ns	1.00
array/accumulate/Int64/dims=1	83759 ns	83529 ns	1.00
array/accumulate/Int64/dims=2	158398 ns	157818 ns	1.00
array/accumulate/Int64/dims=1L	1709802.5 ns	1709864 ns	1.00
array/accumulate/Int64/dims=2L	966551 ns	966626.5 ns	1.00
array/accumulate/Float32/1d	109419 ns	109404 ns	1.00
array/accumulate/Float32/dims=1	80920 ns	80482 ns	1.01
array/accumulate/Float32/dims=2	148065.5 ns	147930.5 ns	1.00
array/accumulate/Float32/dims=1L	1618777 ns	1618960.5 ns	1.00
array/accumulate/Float32/dims=2L	698803 ns	698784 ns	1.00
array/construct	1258.7 ns	1292.1 ns	0.97
array/random/randn/Float32	48210 ns	45521 ns	1.06
array/random/randn!/Float32	24868 ns	24996 ns	0.99
array/random/rand!/Int64	27329 ns	27279 ns	1.00
array/random/rand!/Float32	8602.833333333332 ns	8755 ns	0.98
array/random/rand/Int64	38111 ns	30194 ns	1.26
array/random/rand/Float32	13226 ns	13350 ns	0.99
array/permutedims/4d	60712.5 ns	59834 ns	1.01
array/permutedims/2d	54395 ns	53788.5 ns	1.01
array/permutedims/3d	55384.5 ns	54729 ns	1.01
array/sorting/1d	2758863 ns	2758146 ns	1.00
array/sorting/by	3345133 ns	3344254 ns	1.00
array/sorting/2d	1082485 ns	1080860 ns	1.00
cuda/synchronization/stream/auto	1031.3 ns	1008.3 ns	1.02
cuda/synchronization/stream/nonblocking	8114.8 ns	7614.8 ns	1.07
cuda/synchronization/stream/blocking	793.81 ns	789.4040404040404 ns	1.01
cuda/synchronization/context/auto	1194.4 ns	1182.6 ns	1.01
cuda/synchronization/context/nonblocking	7387.2 ns	8392.2 ns	0.88
cuda/synchronization/context/blocking	904.5227272727273 ns	890.468085106383 ns	1.02

This comment was automatically generated by workflow using github-action-benchmark.

[maleadt]

We should probably make ndevices return Int; it's unexpected for length(::Iterator) to return a non-Integer.

[maleadt]

CI failures unrelated.