Add dockerfile and update pytorch-lightning

[ss32]

Adds a working Dockerfile.

Example usage:

docker run --rm -it --gpus all\ 
-v $(pwd)/outputs:/outputs diffusion:latest\
 python scripts/txt2img.py --prompt "red water bottle floating in a pool"\
 --ddim_eta 0.0 --n_samples 1 --n_iter 1\
 --scale 5.0  --ddim_steps 50 --W 512 --H 512\
 --outdir /outputs

Update to conda env file closes NVIDIA/nvidia-docker#49

[99991]

NVIDIA seems to have broken their docker images, resulting in the following error message:

W: GPG error: https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64  InRelease: The following signatures couldn't be verified because the public key is not available: NO_PUBKEY A4B469963BF863CC                                                    
E: The repository 'https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64  InRelease' is not signed.

Contxt:

• GPG error "public key is not available" in Ubuntu 20.04 CUDA 11.4.0 image while building NVIDIA/nvidia-container-toolkit#257
• https://gitlab.com/nvidia/container-images/cuda/-/issues/158#note_934064306

I replaced the first line of the Dockerfile with

FROM nvidia/cuda:11.6.0-base-ubuntu20.04

which made it work again.