Graph GA "reproduce" operations seem to modify parent mol objects

[AustinT]

After playing around with this code as a starting point for my own GA algorithm, I think that the reproduce function of the graph GA modifies the parent mol objects, so that the parent mols will actually change over time, which shouldn't happen. Here are the lines where the bug happens, although I'm not sure exactly which part has the error:

guacamol_baselines/graph_ga/goal_directed_generation.py

Lines 57 to 59 in 98463a7

	new_child = co.crossover(parent_a, parent_b)
	if new_child is not None:
	new_child = mu.mutate(new_child, mutation_rate)

As a simple fix, for my own purposes I'm storing the SMILES strings instead of the mol objects, but as my code is a heavily modified version of the graph GA algorithm here I can't be bothered to make a PR at the moment. Thought I would just post this here if anybody encounters a similar issue in the future.