Raising Exception for Missing mean_tests or std_tests in MAE Metric

[bharath03-a]

Description

This PR addresses issue #1627. Previously, passing tests=[...] to MAE() silently ignored the test, leading to unexpected results in reports. This could confuse users, especially when writing assertions based on the number of tests executed.

Fix

• Added a check in the MAE class constructor to raise a ValueError if tests is provided instead of mean_tests or std_tests.
• The exception clearly instructs the user to use mean_tests or std_tests depending on what they want to test.

I had considered using a pydantic.Config class with extra = "forbid" to prevent accidental usage. It felt like a cleaner and more generalized solution for validating input.

However, I wasn't entirely sure if this would break other internal logic or dynamic field handling elsewhere in the codebase.

If this kind of config-based validation is acceptable or preferred, I'm happy to explore replacing the current manual ValueError check with a more declarative Pydantic approach. Let me know what you think!

[mike0sv]

@bharath03-a I'd say pydantic config approach is preferred if it does not break anything =) maybe we can try to add this for Metric model and see if any test or example fail?

[bharath03-a]

@mike0sv yeah, I agree too adding pydantic config makes sense. Should I go ahead try that out, since this PR is already merged?

[mike0sv]

@bharath03-a Yep that would be great!