chore: build & publish multi-platform image(linux/arm64, linux/adm64) Docker images

[davidgamez]

Summary:

Alternative solution for #1367. This PR addresses the issue related with no supported multi-platform docker with load parameter (PR comment, docker/buildx#59).

Core changes

• Replace gradle:7-jdk11-alpine docker builder image with eclipse-temurin:11 as gradle:7-jdk11-alpine don't support arm64 architecture.
• Create Github matrix to build and test each docker platform
• Create a separate docker publish step without load parameter to publish a multi-platform docker image

Expected behavior:

A multi-platform docker image is published.

Please make sure these boxes are checked before submitting your pull request - thanks!

• Run the unit tests with gradle test to make sure you didn't break anything
• Format the title like "feat: [new feature short description]". Title must follow the Conventional Commit Specification(https://www.conventionalcommits.org/en/v1.0.0/).
• Linked all relevant issues
• Include screenshot(s) showing how this pull request works and fixes the issue(s)

[davidgamez]

I tested in OS and Linux, and I was able to pull the docker image with the right architecture locally. Used:

docker run --rm ghcr.io/mobilitydata/gtfs-validator:latest-test-multiplat --help
docker inspect ghcr.io/mobilitydata/gtfs-validator:latest-test-multiplat

Working Github action example https://github.com/MobilityData/gtfs-validator/actions/runs/5464582100

[github-actions]

✅ Rule acceptance tests passed.
New Errors: 0 out of 1436 datasets (~0%) are invalid due to code change, which is less than the provided threshold of 1%.
Dropped Errors: 0 out of 1436 datasets (~0%) are invalid due to code change, which is less than the provided threshold of 1%.
New Warnings: 0 out of 1436 datasets (~0%) are invalid due to code change, which is less than the provided threshold of 1%.
Dropped Warnings: 0 out of 1436 datasets (~0%) are invalid due to code change, which is less than the provided threshold of 1%.
0 out of 1436 sources (~0 %) are corrupted.
Commit: f83c557
Download the full acceptance test report here (report will disappear after 90 days).
✅ Rule acceptance tests passed.

[github-actions]

✅ Rule acceptance tests passed.
New Errors: 0 out of 1439 datasets (~0%) are invalid due to code change, which is less than the provided threshold of 1%.
Dropped Errors: 0 out of 1439 datasets (~0%) are invalid due to code change, which is less than the provided threshold of 1%.
New Warnings: 0 out of 1439 datasets (~0%) are invalid due to code change, which is less than the provided threshold of 1%.
Dropped Warnings: 0 out of 1439 datasets (~0%) are invalid due to code change, which is less than the provided threshold of 1%.
0 out of 1439 sources (~0 %) are corrupted.
Commit: 8bce645
Download the full acceptance test report here (report will disappear after 90 days).
✅ Rule acceptance tests passed.

[jcpitre]

::set-output is deprecated.
Here is the suggestion for replacement: https://github.blog/changelog/2022-10-11-github-actions-deprecating-save-state-and-set-output-commands/
But they say they postponed removal because of significant use. So I guess it's up to you to change it or not.

[jcpitre]

Can you explain why it seems to build an image twice, one here and once in "Push Docker container image" below. The two command lines to build the images are identical (obtained form the logs) except for the --push

[jcpitre]

There was a problem running this action (see https://github.com/MobilityData/gtfs-validator/actions/runs/5833255963)
I suspect it's because with the matrix we build 2 images but give them the same tag (ghcr.io/mobilitydata/gtfs-validator:latest, see line 95 below).
Since it appears the matrix strategy runs in parallel for the 2 platforms, this could cause a race condition and it will fail or not depending on the time it takes to create and put the images in the local registry.
I suggest you add a suffix to the DOCKER_TAGS variable on line 95 below. Something like:
DOCKER_TAGS="${DOCKER_IMAGE}.${{ matrix.platform }}:latest"
And see what happens.

[jcpitre]

Tried it and the problem is still present. Although I think it would not hurt to keep this fix.

[jcpitre]

We need to run the arm64 version on an arm64 platform, or use emulation.