chore!: add containerization and packaging manifest lints

[justinthelaw]

Description

Adds linting for Helm, Docker, Zarf, and UDS manifests to improve the quality and consistency of all containerization and Kubernetes related logic.

Adds workflows and pre-commits to ensure consistency before and after committing to a PR.

Remaining lower-priority Dockerfile lint errors are in issue #984

BREAKING CHANGES

• GPU_CLASS_NAME is now GPU_RUNTIME to better align with the actual manifest spec field and purpose
  • RATIONALE: chore!: add containerization and packaging manifest lints #937 (comment)

CHANGES

• Adds Dockerfile linitng
• Adds Helm manifest linting
• Adds UDS bundle linting
• Adds Zarf package linting
  • Moves Zarf variables to override values files (upstream-values.yaml)
• Centralizes linting schemas and configs at the root of the project
  • UDS and Zarf lints rely on a downloaded schema from upstream repository uds-cli
  • Pre-commit checks for and locally downloads for the schema, schema is gitignored
  • Workflow downloads the schema file fresh every time
  • Schemas are pinned to our UDS CLI version of v0.14.0
  • Dockerfile linting config is defined in the .hadolint.yaml
  • Helm linting uses the helm binary lint
• Refactors applicable manifests based on linting errors and warnings
  • "Warning" and "Info" level issues are considered "passing", but can be viewed as outputs in workflow logs (issue chore: burndown remaining dockerfile lint issues #984)
  • All component Chart packages have values.yaml with default values, for UDS overrides following UDS Core patterns
  • A portion of the Dockerfile changes from feat(vllm): upgrade vllm and expose more params for bfloat16 quant compatibility #835 are integrated here to remove linting warnings, and reduce unnecessary layers and dependencies
  • Moves env variables to values files
  • Moves basic pod level configurations to values files
  • Standardizes existing/basic values across all charts
  • Removes autoscaling values that are not tied to a metrics-server implementation nor HorizontalPodScaling resource definition
• Removes GPU_ENABLED python code/env from llama-cpp-python
  • CPU-only package and image (no CUDA deps)
  • No exposure or setting of GPU_ENABLED via values nor Zarf variables
  • Was previously always set to 0 anyway, with no kosher way to change it
• Adds docs related to UDS bundle overrides based on the new values files
• Deactivates e2e_registry1_weekly test on draft PRs (mostly to stop it from running here all the time)
• Removes duplicate check-merge-conflict pre-commit
• More comprehensive clean Make target for build artifact cleaning
• Updates the Registry1 weekly to 0.12.2

Related Issue

Fixes #965 #706 #705 #367 #365

Checklist before merging

• Tests, documentation, ADR added or updated as needed
• Followed the Contributor Guide Steps

[netlify]

✅ Deploy Preview for leapfrogai-docs canceled.

Name	Link
🔨 Latest commit	71476f4
🔍 Latest deploy log	https://app.netlify.com/sites/leapfrogai-docs/deploys/66e09c49a8cf7400083adf6c

[justinthelaw]

Instead of using an outdated and/or sketchy helm-lint action use this to manually install helm v3.15.1 and manually perform a helm lint [sub-directory].

[justinthelaw]

Blocked by #920

[jalling97]

Awesome job cleaning things up!

[vanakema]

some awesome cleanup, ty justin. I didn't review the code closely, but thanks for giving the thorough changelog. This is all great stuff in making the "ux" of contributing and deploying lfai better

All changes completed and explanations reviewed by others

[justinthelaw]

Merge this one in first: #977

Then merge with main and change e2e registry1 weekly to point to the commit SHA as a result of the above merge.