conda and cuda version alignment

[jabowery]

As someone who relies on conda for version alignment, I needed to create an environment.yaml. I also discovered the bash run.sh would fail on a compile error unless I upgraded to cuda 12:

name: CCNeRF
channels:
  - default
  - conda-forge
dependencies:
  - kornia
  - tensorboard
  - torchvision
  - configargparse
  - pytorch
  - ninja
  - numpy 
  - pandas
  - tqdm
  - matplotlib
  - scipy
  - scikit-image
  - plyfile
  - pip:
    - opencv-python