Update identification of BaseFormat during benchmark.yml run

[briangow]

When a new / updated format is pushed to the repo we run a test from benchmark.yml. This test intends to exclude the base formats from the test by searching for Base at the start of the class name. However, not all of the base formats start with Base. All of the base formats are subclasses of BaseFormat though.

This PR searches for BaseFormat in the object bases instead of a class name starting with Base.

[tompollard]

In the guidelines at: https://github.com/chorus-ai/chorus_waveform/blob/main/BENCHMARK.md we say:

If you are defining additional base classes that are not intended to be benchmarked directly, you should include Base at the start of the name (for example, "BaseWFDBFormat"). This allows us to skip the class in the benchmark workflow.

If we are following this approach, then ideally any base formats (i.e. formats that we do not expect to be run directly) should include Base at the start of the name.

The problem with excluding all classes that inherit from BaseFormat is that this will exclude some classes that should be run. e.g.: https://github.com/chorus-ai/chorus_waveform/blob/main/waveform_benchmark/formats/vital.py

[briangow]

Ok, fair enough, I'll close this and update the formats that are missing "Base" at the start of a base class that shouldn't be run.

[tompollard]

An alternative is that we add an explicit flag that determines whether or not the benchmark tests should be run.