Readme improvements

[AnesBenmerzoug]

After conducting user interviews, we realized that the current readme has some issues and that those should be resolved because it is the first point of contact for new users.

General

• Add plots and images to README #336
• Decrease number of steps that lead to first example.
• Put most important content at the top.
• Use absolute URLs for images in order for them to render properly on pypi. #515

List of Papers

• Move list of papers to bottom of readme?
• Move list of papers to documentation #514

Examples

• Fix import issue in README.md example #457
• Consider putting the examples or a more detailed version as colab notebooks. #513

Influence Functions

• Make it explicit that this only works with PyTorch for now.
• Installing pyDVL is not enough. Missing instructions for installing influences extras. #516
• Explain what the return values represent (what are influences?)
• Add printing or plotting of results.

Data Valuation - Shapley Values

• Clarify distinction between shapley values and influence functions. (Done in documentation)
• Explain meaning of Utility to avoid confusion. (Done in documentation)
• Explain what the return values represent (what are values?)
• Explain that data valuation does not only work with sklearn models but rather with all models that implement the sklearn Predictor interface. #533
• Explain TMCS or at least link to documentation part that explains it.

[AnesBenmerzoug]

I am closing this in favour of the smaller issues.