chayma_updates: demos, initialization, mmd estimator, datasets

[Chaymae0612]

Main changes:

• Renamed the demo notebook (demo --> 01_qcbm_training_and_benchmarks)
• Added a second demo notebook for initialization and kernel comparison.
• Updated the README (by adding my name).
• Added data preparation scripts for pairwise and block-pairwise datasets.
• Updated initialization in helpers.
• Updated MMD estimator in models.

[alexiasalavrakos]

Thank you Chayma, it looks great!

Just a couple of comments:

• could you create a new kernels.py file (or maybe kernel_sampling.py? whichever name you prefer) which should be inside src/helpers/ and move the kernel-relevant functions from models/mmd_estimators.py to the new file? And also move the p_sigma function from utils.py, it will fit better in the new kernels.py file/. Thanks!
• in models/mmd_estimators.py can you remove MMD_loss_classical_dataset and MMD_loss_alternative for now (unused)?
• can you write at the end of the new notebook (for example in a markdown cell at the very end) something like "Please note that the demo is set to only 2 iterations per optimization so it can be run fast as an example. This is why we see values of the MMD so close to the random benchmark in the above figure."