Pin conda version for Docker builds (pytorch#164579)

Pin conda version for Docker builds (pytorch#164575)

Mitigates pytorch#164574
Remove unused CUDA_CHANNEL var - this was used before when we had  pytorch install via conda.

Please note: CUDA 13.0 failures are expected since the CI tries to build against prod and CUDA 13.0 is not available in prod yet.
Pull Request resolved: pytorch#164575
Approved by: https://github.com/malfet, https://github.com/Camyll

(cherry picked from commit e40fe63)

Co-authored-by: atalman <[email protected]>

Author: pytorchbot

Dockerfile

@@ -50,11 +50,10 @@ RUN git submodule update --init --recursive
 FROM conda as conda-installs
 ARG PYTHON_VERSION=3.11
 ARG CUDA_PATH=cu121
-ARG CUDA_CHANNEL=nvidia
 ARG INSTALL_CHANNEL=whl/nightly
 # Automatically set by buildx
-RUN /opt/conda/bin/conda update -y -n base -c defaults conda
-RUN /opt/conda/bin/conda install -y python=${PYTHON_VERSION}
+# pinning version of conda here see: https://github.com/pytorch/pytorch/issues/164574
+RUN /opt/conda/bin/conda install -c "${INSTALL_CHANNEL}" -y python=${PYTHON_VERSION} conda=25.7.0
 
 ARG TARGETPLATFORM