Maboss

The generated templates (option -t) were only compatible with version > 2.3.0, this fix makes it compatible.

More portables PRNGs :
- Own implementation of rand48 and glibc rand, added Mersenne twister from <random>

New engines :
- FinalStateSimulationEngine : Only export final state of the simulation
- StochasticSimulationEngine : Only simulate one trajectory, and return only final state

First version of the Python bindings using Python C API
- Simple simulations, with optional outputs
- MaBEstEngine and FinalStateSimulationEngine are available

New objects structure
- Different network, config can be loaded in memory, at the same time

Continuous integration for testing and deploying
- Testing classic and ensemble engine
- Testing server
- Testing with > 128 nodes
- Testing different PRNGs
- Testing server in Docker and Singularity
- Testing on linux and macosx

- Deploying server docker to DockerHub
- Deploying conda libraries
- Deploying cmaboss to pypi

Initial conditions are now a property of a network.
This allows to simulate an ensemble of model, where each model has its own initial conditions.

- Added support for Observed State Transition Graphs, a new functionality introduced in MaBoSS 2.6.0
- Improved Lazy loading support for very large PopMaBoSS simulations

- Initial support of PopMaBoSS, using cmaboss or existing simulation files.
- Updated version of cMaBoSS (v1.0.0b24), which comes with numpy 2.0 support and memory improvements on linux
- pipelines.filter_sensitivity not using strict inequalities.
- added option to get last_* data as a series (as_series=True), for now defaulting to False.