tests: Use pytest as a test runner for Jupyter and Python samples

[kdestin]

Description

This pull request adds initial support for using pytest to run Jupyter Notebooks to facilitate sample validation.

Specifically this pull request:

• Adds infrastructure to enable testing of samples
  • Test resource deployment with bicep
  • pytest as a frontend for running sample validation
    • Custom plugin for automatic discovery of samples as "pytest tests" and executes them with papermill
    • Custom plugin that only runs samples that have changed in a pull request
• Fixes some broken evaluate notebooks

Background

Azure/azureml-examples tests its samples using a github actions as a test runner. Maintaining this prior art revealed some pain points:

• Difficult to orchestrate validation runs (everything runs in parallel all at once, with limited options for control).
• Hard to run a non-trivial number of samples locally.
• Monitoring story isn't ideal. GitHub's UI for actions isn't optimized for repos with hundreds of workflows, which.
• Onboarding new samples into the test suite is manual and often isn't done by contributors (can be unclear whether untested samples are skipped intentionally).

Additionally, infrastructure deployment is based on a large collection of bash scripts, which can make reasoning about the resources that get deployed difficult.

---

Using pytest as a test runner for azureai-samples addresses several of those pain points:

• The testing workflow is local-first. (Deploy resources with bicep, then run pytest).
• Samples (specifically Jupyter Notebooks and Python samples with included tests) are automatically discovered, and must be manually opted out.
• Test run orchestration can be controlled (sequential by default, configurable with plugins like pytest-xdist, pytest-retry, pytest-randomly, etc...)
• Native support for generating test reports, in a format widely understood by other tools (junit xml reports)

Bicep for deployment should make maintaining infrastructure deployments easier (rule of least power).

Checklist

• I have read the contribution guidelines
• I have coordinated with the docs team (mldocs@microsoft.com) if this PR deletes files or changes any file names or file extensions.
• This notebook or file is added to the CODEOWNERS file, pointing to the author or the author's team.