No, it's clear that the researchers are using the term trapdoor in a technical sense. Specifically, one of the key ideas in the swiss system is that it produce a proof that the votes produced as the result of the "shuffle" operation have the same meaning as the votes that are provided as inputs. The easiest "proof" of this would simply be to publish the input votes... but that would defeat the whole purpose of the shuffling. Instead, the Swiss system appears to involve a trapdoor commitment that produces a non-reversible token (hence "trapdoor") that could only be generated if this is a legitimate shuffling. As I read it, this is much like (for instance) a SHA-256 hash of an input message that can be generated to ensure that a document has not been tampered with.
