Ippm denoiser optimisations

[anirudh1666]

Pretty big overhaul of denoiser.py. Key things:

• Max Pooler O(nlogn + n) -> O(n). Came up with a constant time scheme to map points to bins.
• Adaptive Max Pooler same as above
• Readability: Refactored a lot of code to make it more readable, simpler, and easier to maintain.
• Decoupling: Decoupled a lot of code, so you can plug in different variants easier. For instance, to add a new sk-learn clustering algorithm, all you need to do is create the class and initialise the clusterer with the parameters
• Updated tests to reflect new package structure. A note on testing strategy, there are a lot of tests for custom clusterers because AdaptiveMaxPooler was a bit tricky, so I wanted to test all of the edge cases.

Apologies for the long CR, I will make future ones smaller, so it is not as much of a pain to review them.

CRs up-coming:

• Do some more refactoring to make it a bit more consistent (reading over the CR now, I can see some inconsistencies and redundancies, e.g., LabelInfo, calling labels bins, etc.
• Add documentation to new code.
• Refactor IPPMBuilder and Plotter to conform to new code (readability + efficiency)
• Add documentation to builder + plotter
• If time, improve test suite, especially integration tests for customer endpoints, e.g., DenoisingStrategy.fit

[anirudh1666]

Since ruff was failing, I fixed inconsistencies and redundancies. Apart from documentation and more tests, I think this portion of the code is good to go. Also, I used git commit --amend to try and amend my previous commits instead of adding duplicate ones, leading to a polluted commit history. Am I doing this wrong or is maybe the config set-up differently?