Build CUDA packages on CPU agent, test on GPU agent

[h-vetinari]

We're currently building on

pytorch-cpu-feedstock/recipe/conda_build_config.yaml

Lines 12 to 13 in 4d13808

	github_actions_labels: # [linux]
	- cirun-openstack-gpu-2xlarge # [linux]

There are not many physical GPUs available, so this creates a big bottleneck. It would be nice to build the package on a CPU agent (the vast majority of the time for a given run), and then use a GPU-enabled agent only for the testing part.

This idea has already been on the table for a long time, but needs support (at least) in smithy: conda-forge/conda-smithy#1472.

Opening for visibility, since pytorch is most affected currently by this bottleneck (AFAIU, there's also still a physically broken & offline GPU, which further exacerbates the problem).

[mgorny]

Hmm, thinking about it, perhaps we could move CPU-only builds to the regular builders?