fix: add additional cleanup in docker publish workflow, running out of runner space

[lawrence-u10d]

Same code applied to the CI workflow in a previous PR but it's required on the docker publish workflow as well due to the limited github runner disk space

https://github.com/Unstructured-IO/unstructured-api/blob/main/.github/workflows/ci.yml#L120-L140

---

Note

Introduces proactive disk cleanup to prevent runner out-of-space failures during image builds.

• Adds Free up disk space step in docker-publish.yml that removes large preinstalled toolchains (e.g., dotnet, ghc, boost, Android, CodeQL, ghcup, Swift) and runs docker system prune -af --volumes, logging df -h before/after
• Keeps the existing image build logic intact, now executed in a subsequent Build image step

^{Written by Cursor Bugbot for commit 20d2aef. This will update automatically on new commits. Configure here.}