CI cache and linux-arm

[chizkiyahu]

1. add running on ubuntu-24.04-arm
2. add docker build cache on ghcr.io reduce rebuild to < 1 minute
3. docker build add ci_build.sh to repo sync and build code this reduce the build-with-manifest.yml time by >50%
4. weekly rebuild docker without ghcr.io cache
5. replace arm in ${{ github.repository_owner }} to allow easy work in fork repos
6. fix trigger-manifest-build.yml

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chizkiyahu]

@nlithammer @davidegrohmann
Regarding point 5, "replace arm in ${{ github.repository_owner }} to allow easy work in forked repos":
This change helps me with CI development, but it also means that anyone who forks ai-ml-sdk-for-vulkan and runs the actions will also need to fork ai-ml-sdk-manifest.

I’d love to hear your thoughts on this and get as much feedback as possible on the PR.

[nlithammer]

Nice work! You are kind of jumping ahead of what we have ongoing 😄 I have a fix coming up for the collect workflow, but it should be independent of this. I like the new build script, it was on my todo list as well.

[davidegrohmann]

Hi @chizkiyahu thanks for this PR. Great work! There are many things in here that are of interest for us.
May I suggest to split this PR in several smaller PRs for easier reviewing and merging?
For example, we can have the building for ubuntu-arm docker in a separate patch, same for the refactor that extracts the build script in a separate file, and finally extending the build systest to build on arm in a third PR.

[davidegrohmann]

@nlithammer @davidegrohmann Regarding point 5, "replace arm in ${{ github.repository_owner }} to allow easy work in forked repos": This change helps me with CI development, but it also means that anyone who forks ai-ml-sdk-for-vulkan and runs the actions will also need to fork ai-ml-sdk-manifest.

I’d love to hear your thoughts on this and get as much feedback as possible on the PR.

I think that is kinda expected that you need to fork the manifest repo if you want to run this workflow on your own fork. So I see no problems with that. Have you found any issues with this approach?

[nlithammer]

These change of folders, is it related to the refactored Dockerfile? I would prefer to keep the sdk folder.

[chizkiyahu]

I push a fix

[chizkiyahu]

Hi @chizkiyahu thanks for this PR. Great work! There are many things in here that are of interest for us. May I suggest to split this PR in several smaller PRs for easier reviewing and merging? For example, we can have the building for ubuntu-arm docker in a separate patch, same for the refactor that extracts the build script in a separate file, and finally extending the build systest to build on arm in a third PR.

I am converting this PR to draft and opening new PRs

[chizkiyahu]

@davidegrohmann
I split this PR to

• Enhance Docker image build #64
• ci_build script #65
• trigger-manifest-build fix permissions #66
• CI arm #67
• Add .github/README.md #68