bug fixes

[perib]

fixed an issue where the old_params in the individual's NodeLabel were not being reset when a new method was selected. Previous solution tried to update the parameters of the old node. This version creates an entirely new node and replaces the old node in the graph.

Also fixed an issue where we could not set both inner_config_dict and leaf_config_dict to None. When mutate_insert_leaf tried to add a leaf, an error was thrown. Now this mutation function is not a possible option unless one of the inner or leaf config dictionaries is set.

I tested this with a configuration dictionary that includes only xgboost and sgdcclassifier since they share a parameter name with different types, which would cause errors in the hyperparamter update function.

from xgboost import XGBClassifier
from sklearn.linear_model import SGDClassifier
from tpot2.config.classifiers import params_SGDClassifier, params_XGBClassifier

root_config_dict =  {SGDClassifier : params_SGDClassifier,
                     XGBClassifier : params_XGBClassifier,}