wip: vmm_tests: shard vmm tests across multiple runners

[justus-camp-microsoft]

No description provided.

[Copilot]

Pull request overview

This PR implements test sharding for VMM tests to distribute test execution across multiple CI runners, reducing overall test execution time.

Changes:

• Introduces NextestPartition struct to represent test shard configuration (shard number and total shards)
• Threads the partition parameter through the entire test execution pipeline (from job configuration to nextest command generation)
• Configures specific test jobs to run in 3 shards: x64-windows-intel and x64-windows-amd

Reviewed changes

Copilot reviewed 11 out of 14 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
flowey/flowey_lib_common/src/gen_cargo_nextest_run_cmd.rs	Adds NextestPartition struct and generates --partition count:shard/total argument for nextest
flowey/flowey_lib_common/src/run_cargo_nextest_run.rs	Threads nextest_partition parameter through the run pipeline
flowey/flowey_lib_hvlite/src/test_nextest_vmm_tests_archive.rs	Adds partition parameter to test archive execution
flowey/flowey_lib_hvlite/src/run_cargo_nextest_run.rs	Passes partition parameter to common library
flowey/flowey_lib_hvlite/src/build_nextest_vmm_tests.rs	Sets partition to None for build-time test execution
flowey/flowey_lib_hvlite/src/build_nextest_unit_tests.rs	Sets partition to None for unit test builds
flowey/flowey_lib_hvlite/src/test_nextest_unit_tests_archive.rs	Sets partition to None for unit test archives
flowey/flowey_lib_hvlite/src/_jobs/local_build_and_run_nextest_vmm_tests.rs	Sets partition to None for local builds
flowey/flowey_lib_hvlite/src/_jobs/consume_and_test_nextest_vmm_tests_archive.rs	Adds partition parameter and passes it through to test execution
flowey/flowey_hvlite/src/pipelines/checkin_gates.rs	Configures shard_count for specific test jobs and generates multiple jobs per shard
ci-flowey/openvmm-pr.yaml	Generated ADO pipeline with new sharded test jobs (job16-job18 for intel, etc.)
.github/workflows/openvmm-ci.yaml	Generated GitHub Actions workflow with new sharded test jobs

[smmalis37]

What's the difference between None and Some(1)? Why not just make this a plain usize?

[justus-camp-microsoft]

No issue I guess, the nextest invocation would just end up having --partition count:1/1 which is harmless. In any case, I'm not sure sharding this way ends up being worth it from testing

[justus-camp-microsoft]

So I tried 2 shards and 3 shards, here's the result. Looks like 2 shards ends up basically being wall-clock neutral because of how nextest shards the tests across runners. 3 shards gives us ~15 minutes of wall-clock time savings at the cost of a 10% increase in total CI runner time. Does that seem worth it?