Define pipeline_base __copy__() or __deepcopy__() operator

A common way to execute a deepcopy is to use the python builtin copy library. Currently, in the EvalML baseline, we use the `pipeline.clone()` and `pipeline.new()` methods. Does it make more sense to define the `__copy__()` impl and `__deepcopy__()` impl and refactor to import the copy library and do `copy.copy(pipeline)` or `copy.deepcopy(pipeline)` instead?