feat: Updated CI for multi platform builds

[malteos]

This PR enables building container images for linux/amd64 and linux/arm64 platforms. Fixes #27

Verified on MacOS:

$ uname -p
arm

$ podman pull ghcr.io/commoncrawl/cc-crawl-statistics/stats:pr-28 

$ podman inspect ghcr.io/commoncrawl/cc-crawl-statistics/stats:pr-28 | grep Architecture
          "Architecture": "arm64",

[lfoppiano]

QEMU is OK I guess for this case, but the build will be slower than running two parallel jobs on native architectures and merging them as I mention in the issue.

[malteos]

Is build speed an issue? I prefer this because keeps the CI yaml simple but I am happy to change as well.

[malteos]

What does @sebastian-nagel prefer? See for reference: https://github.com/grobidOrg/grobid/blob/b36cdd4fbf33661678dcbf213fe5ec38baa8478e/.github/workflows/ci-build-unstable.yml

[lfoppiano]

I think it depends. 😅

If we keep running the image on our own computer and we might need to change the code if something happens, I'd prefer a faster buidl.
However, I can imagine having this docker image running on rf and provide an interface where we could pull the data and generate the statistics remotely, without having to use our laptop (took at least 30 minutes on my m2).

This would need some more work, but if so we won't care about build time IMHO.

[lfoppiano]

Could you please trigger a build so that I can test also the resulting image? I will try to process the stats for this month to compare the speed with an non-arm version 🙏

[malteos]

Could you please trigger a build so that I can test also the resulting image? I will try to process the stats for this month to compare the speed with an non-arm version 🙏

It's already built: https://github.com/commoncrawl/cc-crawl-statistics/pkgs/container/cc-crawl-statistics%2Fstats/703032387?tag=pr-28