Add ability to install GLT through a post-install-script

[svij-sc]

Scope of work done

Still some work left to be able to run the e2e examples.
Limiting the scope for this just to be able to install GLT as a post-install step. The scope of getting this to work w/ our quick start instructions e2e is much larger - thus want to get in first.

With this change you should be able to:

1. Install keyring so pip can pull packages from google artifact registry

pip install keyrings.google-artifactregistry-auth

2. Install gigl

pip install "gigl[torch25-cuda-121,transform]==0.0.3" \
  --index-url=https://us-central1-python.pkg.dev/external-snap-ci-github-gigl/gigl/simple/ \
  --extra-index-url=https://pypi.org/simple

3. run the post install script to install GLT

gigl-post-install

Where is the documentation for this feature?: N/A

Did you add automated tests or write a test plan?

Updated Changelog.md? NO

Ready for code review?: YES

[svij-sc]

/unit_test

[svij-sc]

/integration_test

[svij-sc]

/e2e_test

[github-actions]

GiGL Automation

@ 18:19:37UTC : 🔄 Unit Test started.

@ 19:00:13UTC : ✅ Workflow completed successfully.

[github-actions]

GiGL Automation

@ 18:19:40UTC : 🔄 Integration Test started.

[github-actions]

GiGL Automation

@ 18:19:46UTC : 🔄 E2E Test started.

@ 19:52:05UTC : ✅ Workflow completed successfully.

[github-actions]

GiGL Automation

@ 20:42:29UTC : Starting to build base images for CUDA and CPU.
This may take a while, please be patient.
The images will be pushed to the GCP Artifact Registry.
Once done, the workflow will update the dep_vars.env file with the new image names.

[github-actions]

GiGL Automation

@ 21:02:46UTC : Starting to build base images for CUDA and CPU.
This may take a while, please be patient.
The images will be pushed to the GCP Artifact Registry.
Once done, the workflow will update the dep_vars.env file with the new image names.

[nshah-sc]

I'm fine with this solution -- I agree there might be another path but ultimately I think this is a pretty sufficient unblock to current issues. Compiling the wheels to simplify even further would be nice but I don't think the current solution is overly complex or error-prone so I don't see a too-strong need to carve out additional work to support that.

[github-actions]

GiGL Automation

@ 22:50:03UTC : Starting to build base images for CUDA and CPU.
This may take a while, please be patient.
The images will be pushed to the GCP Artifact Registry.
Once done, the workflow will update the dep_vars.env file with the new image names.

[github-actions]

GiGL Automation

@ 21:09:35UTC : Starting to build base images for CUDA and CPU.
This may take a while, please be patient.
The images will be pushed to the GCP Artifact Registry.
Once done, the workflow will update the dep_vars.env file with the new image names.

[github-actions]

GiGL Automation

@ 21:27:20UTC : Built and pushed new images:

• CUDA base image: us-central1-docker.pkg.dev/external-snap-ci-github-gigl/public-gigl/gigl-cuda-base:148f1c215806a7feb36663180c6e200548a94da4.25.1
• CPU base image: us-central1-docker.pkg.dev/external-snap-ci-github-gigl/public-gigl/gigl-cpu-base:148f1c215806a7feb36663180c6e200548a94da4.25.1
• Dataflow base image: us-central1-docker.pkg.dev/external-snap-ci-github-gigl/public-gigl/gigl-dataflow-base:148f1c215806a7feb36663180c6e200548a94da4.25.1
• Builder image: us-central1-docker.pkg.dev/external-snap-ci-github-gigl/gigl-base-images/gigl-builder:148f1c215806a7feb36663180c6e200548a94da4.25.1

Updated dep_vars.env with new image names.
Updated .github/cloud_builder/run_command_on_pr_cloud_build.yaml with new builder image.

[kmontemayor2-sc]

LGTM - would prefer if @yliu2-sc or @nshah-sc stamped too tho

[nshah171]

thanks for the work!