Missing information requests for complete agreements are a bit strange: they're requests for messages that may not have originally been sent between the requester and the requestee; they don't actually resolve the problem of disagreement regarding whether the doubtful bilateral agreement's payments should be executed; and they provide little useful information beyond what's contained in the partial agreement (which can be requested if necessary).
Furthermore, the current specification regarding requests for complete agreements doesn't unambiguously handle the subtleties of how many different ways an instance can come to the conclusion that a particular complete agreement isn't being executed. (For example, what if every relay receives the complete agreement before the relevant deadline, but they all receive a different second complete agreement before the relevant reversal deadline?) Also, it glosses over the distinction between "on time" meaning "before the relevant deadline" and "on time" meaning "before the relevant reversal deadline".
Perhaps it's better, for the sake of the simplicity of the protocol and its implementations, not to have missing information requests for complete agreements at all.