[MNT] Add testing for loading/saving for deep clusterers

[hadifawaz1999]

Describe the issue

In the deep clustering module there is the possibility of loading a pre-trained model, however the function is not tested, which lead to the bug that was fixed in #3074

Suggest a potential alternative/fix

Add a test for this functionality

Additional context

No response

[satwiksps]

Hi @MatthewMiddlehurst @hadifawaz1999

My plan is to add unit tests for the save/load functionality of the deep clusterers, specifically targeting the models that required the fixes in #3074 (DRNNClusterer and DCNNClusterer).

The test will:

• train each model on a small synthetic dataset,
• save the trained model,
• load it back using the updated load() method, and
• verify that the loaded model is of the correct type and can produce valid predictions.

This should ensure that future changes do not break the custom-module loading behavior again.

[MatthewMiddlehurst]

Before closing this it would be good to consider if the test can be part of the testing/ module rather than its own file.