use filenames to define input biopolymer and ligand files in SetupSampler.from_toml to setup systems

[kntkb]

currently, the biopolymer files are defined using target_class and target_name in config.yaml. This was initially intended to be used for training espaloma with RNA reweighting, but this should be more generalizable for other purposes. For example, create protein-ligand or protein systems using espaloma for pure simulation purposes.