Test generated workflow matrix schemas

[ScottTodd]

Motivation

Overall CI test quality improvement as part of #3336.

We've been replacing inline workflow matrix code with dynamic matrix generator scripts but the data produced by these scripts was only tested by actually running the workflows. This adds unit test coverage ensuring that every matrix.value key in each workflow file is actually produced for every row of a generated matrix. This won't catch all errors, but it will give us more confidence and guard against typos.

See prior PRs:

• Run PyTorch tests in multi-arch release workflows #5661
• [ci] Add optional PyTorch build toggle for workflow_dispatch (re-land) #5727
• Generate dynamic matrix for ROCm python package tests in CI #5942
• Add optional JAX release build toggle #6117

Technical Details

I decided to start by just checking one way: every matrix.value in a workflow must be produced by the generator. The opposite direction (the generator producing a value not used by a workflow) is not checked yet. Some jobs may only want to read a subset of the values and that could be okay.

Test Plan

Each of the new tests can fail with mismatches in either the workflows or the generator scripts:

--- a/.github/workflows/multi_arch_build_windows_pytorch_wheels.yml
+++ b/.github/workflows/multi_arch_build_windows_pytorch_wheels.yml
@@ -473,7 +473,7 @@ jobs:
     uses: ./.github/workflows/test_pytorch_wheels.yml
     with:
       amdgpu_family: ${{ matrix.amdgpu_family }}
-      test_runs_on: ${{ matrix.test_runs_on }}
+      test_runs_on: ${{ matrix.unknown_key }}

--- a/build_tools/github_actions/configure_pytorch_test_matrix.py
+++ b/build_tools/github_actions/configure_pytorch_test_matrix.py
@@ -69,7 +69,7 @@ def build_test_matrix(
         include.append(
             {
                 "amdgpu_family": requested_family,
-                "test_runs_on": test_runs_on,
+                "different_unknown_key": test_runs_on,
             }
         )

with errors like

                for row in matrix["include"]:
                    # This checks the row schema, not whether values are
                    # truthy. Empty values are allowed when a workflow handles
                    # them explicitly. Undefined values are not: if the
                    # workflow reads `matrix.unknown`, this test fails until
                    # the generator emits that key for every row.
>                   self.assertEqual(matrix_references - set(row), set())
E                   AssertionError: Items in the first set but not the second:
E                   'test_runs_on'

build_tools\github_actions\tests\configure_pytorch_test_matrix_test.py:223: AssertionError

Submission Checklist

• Look over the contributing guidelines at https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.

[therock-pr-bot]

❌ PR Check — Action Required

Check	Status	Details
🌿 Branch Name	✅ Pass	—
📝 PR Title/Description	❌ Fail	Error: Title does not follow Conventional Commits style. Expected: start with a valid type (feat, fix, docs, …). Desired format: type(optional-scope): short description
⛔ Forbidden Files	✅ Pass	—
🧪 Unit Test	✅ Pass	—
🔎 pre-commit	✅ Pass	—
🚫 Draft PR	🔜 To Be Enabled	—
🚩 Feature Flag	🔜 To Be Enabled	—
📊 Code Coverage	🔜 To Be Enabled	—

⚠️ 1 policy check(s) failed. Please address the issues above before this PR can be Reviewed.

🚫 Please fix the failed policies

• ❌ PR Title/Description

The Not ready to Review label was added to this PR. Once all policies pass, the label is removed automatically.

📖 Need help? See the Policy FAQ for details on every check and how to fix failures.

[therock-pr-bot]

🚫 Please fix the failed policies before requesting reviews.

The following policy checks failed:

• ❌ PR Title/Description

The Not ready to Review label has been added to this PR.
Once all policies pass, the label will be removed automatically.