`relu` propagates NaN on CPU but not on GPU

edit: see https://github.com/FluxML/NNlib.jl/issues/509, this is due to the `relu`, not really the `BatchNorm`

```julia
julia> using Flux

julia> layer = BatchNorm(32, relu)
BatchNorm(32, relu) # 64 parameters, plus 64 non-trainable

julia> layer(NaN32*zeros(Float32, (32,1)))
32×1 Matrix{Float32}:
NaN
NaN
NaN
NaN
NaN
NaN
NaN
⋮
NaN
NaN
NaN
NaN
NaN
NaN

julia> gpu(layer)(gpu(NaN32*zeros(Float32, (32,1))))
32×1 CuArray{Float32, 2, CUDA.Mem.DeviceBuffer}:
0.0
0.0
0.0
0.0
0.0
0.0
0.0
⋮
0.0
0.0
0.0
0.0
0.0
0.0
```

edit: just saw I'm on Flux v0.12.10 so maybe this is outdated

Start by reproducing the CPU and GPU examples using BatchNorm(32, relu) with NaN32 input, then inspect the relu implementation and its backend-specific behavior. Done means relu handles NaN consistently on CPU and GPU, with tests covering the reported discrepancy.

Written by the indexing model from the issue text.