ci: run tests periodically on main

[jgehrcke]

Note: GitHub evaluates the schedule trigger only from the workflow file located in the default branch. In other words, scheduled GitHub Actions workflows only run on the repository's default branch.

Multi-faceted motivation.

Running CI periodically reveals unexpected breakage faster, and provides valuable stability distribution data.
Was reminded by #895.

[jgehrcke]

That doesn't necessarily make sense to me:

#12 [9/9] RUN curl -sSfLO --retry 8 --retry-all-errors --connect-timeout 10 --retry-delay 5     https://dl.k8s.io/release/v1.34.0/bin/linux/arm64/kubectl &&     chmod ugo+x kubectl && mv kubectl /usr/bin
#12 CACHED

#13 exporting to image
#13 exporting layers done
#13 exporting manifest sha256:830e9ff2aa384ac1a5dcca1014d37f51b4cda4ba91511a1a763bc617f9841a4e done
#13 exporting config sha256:91f4907a0a167a1a655475ce489bac3fe83ba6c38227e61d73edc41c73a58165 done
#13 exporting attestation manifest sha256:fcf0df8ab6cad52ffa833cdcc49a67c6f2f4251481a62919e7346dc59f68f496 done
#13 exporting manifest list sha256:a9422df60b423e2da247fa4e559c2e546726d52e65ca870b860902c485bb1990 done
#13 naming to docker.io/library/batstests:d7b8018a done
#13 unpacking to docker.io/library/batstests:d7b8018a done
#13 ERROR: failed to extract layer sha256:3de26a0388bd1fba21d203f8109c7ff960d71981c2cfd311e8726886cc748904: failed to get reader from content store: content digest sha256:048e83909d3e5c6859ba35783350c627d432989d27698637b31e2d7884e770da: not found

(https://github.com/NVIDIA/k8s-dra-driver-gpu/actions/runs/22494401585/job/65408030943?pr=896#step:3:50)
This is a local operation that failed. I ran docker builder prune -a locally, and that is likely to have fixed that. Rerunning.

[jgehrcke]

Let's get this in.