Hi, since there is a [PR](https://github.com/JuliaMath/FFTW.jl/pull/222) for `rfft!` in FFTW we probably should also add `rfft!` in AbstractFFTs? In general [in-place real-to-complex FFT](https://docs.nvidia.com/cuda/cufft/index.html#data-layout) seems to be supported by CuFFT as well, right? I recall that I was discussing this somewhere with @maleadt but I can't remember exactly. What do you think about that? Best, Felix