I haven’t; the “us” that you refer have regularly posted blanket claims about polls being wrong, without the elaboration in your premise.
The 12 point lead was an outlier even at the time. 10/23/16! IIRC the polling average - see 538 or realclearpolitics - was 6 points. These sites also show the precipitous drop in that lead in the last ~10 days. The polls opened and closed throughout the campaign. Why would you highlight the result of a particular poll on a particular day and conflate that with a prediction?
You are correct, however, that in addition to the polling statistics, the turnout modelling is an important part of the process. 538 has looked for systematic problems by comparing predictions to results over decades and weighting the various pools according to the quality of their prediction in his averaging. That is why he predicted a closer election that others. Whatever the plusses and minuses of the models, public pollsters need to strive for accuracy; their reputation and livelihood depends on it.
It is interesting that the that in about a third of the paths, Trump wins, yet their statistical modelling had 17:3 odds against. They may have underestimated the correlation in voting patterns, and treated the votes in various states as though they were independent trials. Silver incorporated a model of correlation in his model and that is presumably why his odds were much lower at 2:1. These odds at the day of election did not make Trump’s victory a long shot.