Reject malformed MPIs in v5 packets

This MR addresses part of #143 (closed). It is a subset of !209 (closed), with the aim of being easier to reach consensus on.