[BurhanAbdullah]

Adds SHA-256 manifests for benchmark artifacts, including file sizes, checksums, and optional run metadata. Includes integrity verification, regression tests, documentation, and CI integration.

Validation: Local tests passed; upstream CI is pending.

[kosmylo]

Thanks for the contribution. The checksum manifest has a clear use for detecting missing or modified files in a shared result bundle. However, it is optional artifact-integrity tooling rather than a requirement for benchmark execution or scoring.

I suggest deferring this PR until we define a concrete publication, archival, or submission workflow that will use the manifest. A documented manual workflow would be sufficient, but the current proposal does not establish why the benchmark needs to maintain this additional CLI and file format now.

If we proceed, please reconcile the branch with current main and preserve the scoped Level 2 test command introduced in PR #10. The older workaround that installs pytest for unrestricted unittest discovery is no longer needed for that job.