CI/CD Visual regression testing

[Bombfuse]

ideally we'd be able to run Emerald headless in the CI/CD pipeline, and additionally run examples, screenshot them, then compare them to previous results to guarantee nothing has regressed or changed.

Additionally there needs to be a way to OK different results, in the case that we expect visuals to change