Fix package specifications

[ajschmidt8]

After rapidsai/integration#411 was merged, the build string for our meta-packages changed from cuda11.5* to just cuda11*. This change caused our Docker builds to stop working.

Since we now build with CUDA Enhanced Compatibility, I think it should be fine to drop the =cuda spec from the strings below. However, an alternative would be to include the cudatoolkit spec as shown below, but I don't think that's necessary.

gpuci_mamba_retry install -n rapids \
  "rapids-notebook-env=${RAPIDS_VER}" \
  "cudatoolkit=${CUDA_VER}*" \
  && gpuci_conda_retry remove -y -n rapids --force-remove \
  "rapids-notebook-env=${RAPIDS_VER}*"

@ethyling, any thoughts here?

[jjacobelli]

After rapidsai/integration#411 was merged, the build string for our meta-packages changed from cuda11.5* to just cuda11*. This change caused our Docker builds to stop working.

Since we now build with CUDA Enhanced Compatibility, I think it should be fine to drop the =cuda spec from the strings below. However, an alternative would be to include the cudatoolkit spec as shown below, but I don't think that's necessary.

gpuci_mamba_retry install -n rapids \
  "rapids-notebook-env=${RAPIDS_VER}" \
  "cudatoolkit=${CUDA_VER}*" \
  && gpuci_conda_retry remove -y -n rapids --force-remove \
  "rapids-notebook-env=${RAPIDS_VER}*"

@ethyling, any thoughts here?

I think we can drop it for now, and add it later if needed