Fedbiomed

- Implementation of packaging for the library/framework to publish on the PyPI repository.
- CI/CD: configuration for continues deployment.
- Creation of the `fedbiomed` executable, installed alongside the Fed-BioMed installation.
- Documentation updates following related changes.
- Introduction of the PDM package manager for development.
- Updates to Docker containers to eliminate the use of `miniconda`.
- Introduction of VPN Docker container tests.
- Removal of the `environ` class in favor of the improved `Config` class.
- Secure aggregation implementation for advanced federated optimization functionalities.
- Fixes: 'colab' shell recognition issue 1274

- introduce in-application implementation of Shamir Additive Secret Sharing (ASS) for Joye-Libert secure aggregation
- remove dependency to MP-SPDZ library, using in-app ASS
- introduce node to node messages security for honest but curious scenario to support in-app ASS
- improve node to node messages implementation to use symmetric encryption
- refactor `Message` class, improving API to upper layers
- fix researcher jupyter notebook automatic token transmission not working
- fix LOM secagg researcher side aggregation error for values >= 2**31
- misc give clearer node side error message in case of overflow with LOM secagg

- introduce fast secure aggregation using Low-Overhead Masking (LOM) scheme and in-app Diffie-Hellman key exchange
- add feature for choosing the batch size for testing (validation) facility
- add transfer learning tutorial
- fix make optional potentially insecure consistency check (validation) for secure aggregation
- fix issue with validation when `test_batch_size=1`
- multiple fixes in documentation and tutorials

- redesign `Job` and `Experiment` classes, introducing `federated_workflow` class
- introduce end-to-end and testing facility
- fix, robustify and redesign gRPC communication layer
- add a concrete-ml example
- improve logging facility
- fix SecAgg (`Experiment` with more 4 nodes or more was failing)
- fix quantization process for secure aggregation
- review notebooks. Remove unrelevant ones.
- fix and extend `model` class, by handling buffers layers such as batch normalization layers.
- multiple fixes for breakpoint
- add a [concrete-ml](https://docs.zama.ai/concrete-ml) example

- multiple fixes and improvements for gRPC communications stability
- fix scaffold failure when reloading breakpoint and continue training
- fix breakpoint save failure for DecLearn optimizers
- fix import/export model check for explicit error when not instantiated

- fix crash happening when the tkinter file selector is used in non-interactive mode