Add add_target_to_X helper and check init parameter types in encoders

[solegalli]

Shared changes for the encoder PRs (#1026, #1028, #1029, #1031, #1032).

• Add add_target_to_X(nw_X, y) in feature_engine/encoding/_helper_functions.py. It adds y to X as a column, pairing rows by position, so list, array and series targets work. With pandas, the column takes the index of X. It is used by MeanEncoder, OrdinalEncoder, WoEEncoder and DecisionTreeEncoder.
• MeanEncoder uses the helper.
• Check that string parameters are strings before comparing them to the allowed values: unseen (check_parameter_unseen), missing_values (CategoricalInitMixinNA and _check_param_missing_values) and CountEncoder's encoding_method.
• Fix the get_feature_names_out error message, which printed a literal {input_features}.
• MeanEncoder and CountEncoder tests: init tests at the top, init params assigned correctly, and error messages matched in every pytest.raises.