[REF-2089] Use dill instead of cloudpickle for serialization (#2922)

* Use dill instead of cloudpickle for serialization

* smaller size pickles
* support dynamically defined states
* avoid issues with unpickleable globals

* pyproject: add dill, remove cloudpickle

* poetry.lock: relock dependencies

* Dynamically convert EventHandler to functools.partial

Instead of converting the functions up front and assigning them to the
instance, unbox the function from the EventHandler when it is requested via
__getattribute__. This reduces the size of the per-instance pickle, because
event handler bodies do not need to be included.

* Improve checking for cython_function_or_method

Because pydantic can be installed without cython, only use the workaround in
the case where the BaseModel.validate function is NOT a FunctionType,
indicating it's a cython function.

* Serialize all State subclasses by reference