Suppress warning when using GANs that suggests that CUDA isn't being used

[fealho]

CU-86b0vehd0, Resolve #2052.

This PR suppresses the warning generated in CTGAN: https://github.com/sdv-dev/CTGAN/blob/d132c23a7bf5883cba0d1cb09f6d54aa400c39fb/ctgan/synthesizers/ctgan.py#L43

The warning originates from torch: https://github.com/pytorch/pytorch/blob/f72266eceaf833cd1a44d8e4ac611bc36102539c/aten/src/ATen/cuda/CublasHandlePool.cpp#L135C5-L135C20

[sdv-team]

Task linked: CU-86b0vehd0 SDV - UserWarning when using CTGAN caused by potentially uninitialized CUDA context #2052

[frances-h]

Are there cases when this warning is valid? If it's only popping up in error then it's probably ok to silence it, but maybe not if it does provide a useful warning in some cases.

After looking at the warning a bit more, it seems like it's just warning that it's creating the context in the first place so probably ok to silence it 👍