Magen Avraham (447:5) writes that the issur of nosein ta’am l’pegam by chameitz is a lo plug extension of the issur of chameitz b’mashe’hu. Even though one could distinguish between the two – n”t l’pegam has a bad flavor; mashe’u has no flavor -- the assumption is that since neither gives off a ta’am of issur, the two go hand in hand. The issur mashe’u applies to chometz noksheh – what about the issur of n”t l’pegam? The MG”A is willing to at least entertain the possibility that it does not. The implication is that lo pelug does not mean there is one broad blanket gezeirah covering both cases, but rather than there is an issur mashe’hu that was then extended further through a new sibah called “lo pelug.” That new sibas haissur may be limited to only real chameitz and not extend further.
There is a din that a nursing woman may not remarry for a certain period of time lest she become pregnant and be unable to nurse. The gemara (Kesubos 60b) uses lo pelug to extend this dineven to cases where there is no danger to the baby, e.g. the mother hires a wet nurse, the baby was weaned, or the baby dies c”v. Yet, the gemara tells us that R’ Nachman made an exception and allowed a member of the Reish Galusa’s family to remarry right away because she hired a wet nurse. Tosfos writes that the lo pelug here does not tell us to equate all cases, even where the reason for the din doesn’t apply. Were that the case, no exception would be possible. Rather, the lo plug tells us that the reason for the din should be understood as applying even to the most remotely possible cases. If a woman gives the baby to a wet nurse, that’s not a guarantee of anything – maybe she will change her mind and decide to nurse again herself. Therefore, Chazal said lo plug whether she hired a nurse or not. With respect to the Reish Galusa, their going back on their decision was not even a remote possibility -- it was completely off the table, hence the exception.
I would say the shakla v’teryla of Tosfos hinges on this lo plug chakirah. If lo plug is it’s own sibah, then by definition it serves as a justification for extending the din even where the original reason doesn’t apply – it by definition precludes exceptions. But if lo plug is just a siman that the original gezeirah should be interpreted very broadly, then where the original gezeirah has no chance in the world of coming into play, e.g. the Reish Galusa case, exceptions can apply.
Just to play fair, I should mention Rashi in Kesubos sounds like he understands the lo plug there as the former type – apply the din across the board irrespective of the reason involved – and nonetheless, R’ Nachman made an exception. That throws a monkey-wrench into the works here.
Obviously looking at two cases is just a starting point and to do this right you would have to collect all the lo plug cases and see how they work. Unfortunately, I need to leave that job for someone else to pick up on for now.