get_flow fn throws error when dependencies are not met #485

apparently, get flow tries to deserialize the flow immediately into a model. However, sometimes we want to just download and inspect a flow (e.g., a flow from another sklearn version or a flow from any other workbench) and this seems not possible.