Scml

------------------

* typing update
* removing unnecessary sphinxtesters
* requiring latest stable_baselines
* doc update
* bugfix in scml2020 adapter to run oneshot
* Adding skeleton for the new std simulation
* Adding 2024 utility functions and worlds
* Switching docs to scml.readthedocs.io

------------------

* upgrading CI to python 3.12
* importing observations
* dropping support for python < 3.10
* Completing first draft of the RL environment

----------------

* Moving scml.scml2020.utils to scml.utils (to avoid a circular import in rl/factory)
* This is an initial effort to have RL work for scml. Currently we are disabling the extra checks on observation and
world construction because they lead to failures. This needs to be rectified later.
* Added the concept of a WorldFactory to generate worlds with predefined characteristics from the point of view of the learning agent.
* Now environments, action managers, and observation managers all receive WorldFactory objects and automatically check for compatibility.

----------------

* Upgrading to be compatible with NegMAS 0.10.0
* Adding step_with() to oneshot allowing for single counter-offer set stepping of the simulation (to be used to expose the simulation as RL and MARL environments later).
* Adding current_negotiation_details to OneShotAWI to get details of running negotiation
* Adding managed sales, supplies, total_sales, total_supplies, needed_sales, and needed_supplies to the OneShotAWI
* Extending OneShotState returned from awi.state and awi.default_state_encoder(mechanism_states) to expose more information about the simulation

----------------

* Upgrading to be compatible with NegMAS 0.9.8
* Updating notebooks for 2023 release
* Adding 2023 specific classes and utility functions
* Supporting most recent negmas version on Github
* Adding the ability to use raw collusion scores
In the collusion track, it is now possible to control how much does the
raw collusion score affect the final score relative to the difference
between raw collusion score and standard score.
* more reobust renaming when reveal-names is set

-----------------
* bugfix in anac2022_* methods: Score reporting was not accurate because of a change in naming convention in negmas.
* adding agent-processes to SCML2022World.generate
* Adding SCML2022World which is just an alias for SCML2021World
* bugfix 96 Partial List of Agent Scores: Also discussion $94
* Incorrect agent types and mysterious agents appearing in the scores
list.
* Inaccurate placement of agents in the simulation
* disabling tests taking too long on CI
* commenting members of Profile, Contract in oneshot