Harmonize and clean the repo for stage 2

[LukasHats]

For streamlined processing during stage 2, cleaning and harmonizing the repo is needed.

• (OPTIONAL)Harmonize segmentation mask and image names. Will affect many method running files!
• SDS File structure: Dataset Folder structure; Harmonize_names from methods in results folder
• Harmonize marker names for unified XGBoost
• Fix (relative) pathing on all scripts, including plotting scripts. E.g. TO reproduce plots/results of manuscirpt, deposit the results csv correctly and assess using the plotting scripts @ArozHada
• Harmonize everything around Nimbus and the respective clustering method
• All methods should appear with a subfolder under src. Inside, each method should have 1 executable running file if possible, e.g. run_scyan.py. For clustering methods this needs to be checked specifically @MargotCh
• Have a benchmarking utils subfolder containing e.g. a supervised_utils.py and a foundationa_models_utils.py. REspective models should import functions from there. Same goes for clustering methods.
• Once again check all cell type names and verify harmonization @LukasHats

IMPORTANT!: For each cleaning step, please check which files are affected by it and adjust accordingly!

[ArozHada]

Proposed folder structure

spCellEval/
├── src/                          # Core benchmarking logic
│   ├── collect_results.py        # Main script to aggregate/collect results
│   └── methods/                  # Each method gets its own subfolder
│       ├── method1/
│       │   ├── run.py
│       │   └── ... (scripts, models, configs)
│       ├── method2/
│       └── ...
├── data/
│   ├── raw/
│   └── processed/
├── results/                      # Results produced by methods or aggregation
│   ├── method1/
│   ├── method2/
│   └── summary/
├── website/                      # Website code (HTML, CSS, JS, assets, deploy configs)
│   ├── index.html
│   ├── css/
│   ├── js/
│   └── assets/
├── notebooks/                    # Exploratory or analysis notebooks
├── docs/                         # Documentation (user guide, API, etc.)
├── tests/                        # Automated tests for benchmarking and methods
├── environment.yml               # Dependencies
├── .gitignore
├── LICENSE
└── README.md

[LukasHats]

Yes looks good!. Just the data deposit is problematic regarding upload data size restriction on GitHub. So we will not be able to save per-method results but rather the final_results.csv only. But we could deposit the extended results on the public repository and access it with an API key from the scripts. At least at the final stage 2