Cuda version change in documentation

The command to install Nvidia GPU support on Linux machines was:

$ CONDA_OVERRIDE_CUDA=11.2 mamba install 'pytorch-gpu==cuda112'

and now it has been changed to:

$ CONDA_OVERRIDE_CUDA=11.8 mamba install 'pytorch-gpu==cuda118'

doc/install.rst

@@ -55,7 +55,7 @@ We support two installation modes, through pip_, or mamba_ (conda).
 
          # or, to force the Nvidia CUDA version (environments w/o Nvidia setup):
 
-         $ CONDA_OVERRIDE_CUDA=11.2 mamba install 'pytorch-gpu=*=cuda112*'
+         $ CONDA_OVERRIDE_CUDA=12.0 mamba install 'pytorch-gpu=*=cuda118*'
 
 
 .. _mednet.setup: