That said, I don't believe in analyzing crosstabs or unskewing. People just have to understand that Dems consistently outperform their polls in the SW, we just are only used to it in NV because that is the state polled repeatedly for 20 years. The only way to correct for this is to poll in Spanish (preferably using Latinos rather than Anglos with training or Bots) and run your poll 24 hours a day to capture menial workers - two things that cost exponentially more than desirable.
Agreed that we should not just reflexively unskew it.
NYT/Siena does actually poll in Spanish, or at least they did in 2018. However, even when you poll in Spanish I have noticed over time that polls of Hispanic voters in particular tend to lowball Dems. The same is true of exit polls.
It also quite often seems to be the case that if you compare precinct results and either polls or exit polls, Dems tend to do better in terms of actual votes cast than one would expect based off of polls and exit polls of Hispanic voters. Yes, some of that is due to Hispanic voters in heavily Hispanic areas voting differently from ones in other precincts, but the effect normally seems to be larger than is easily explained by that alone.
I do find it plausible that Trump could do at least reasonably well in South TX (and to some degree West TX), especially in the more rural counties; there does seem to be at least somewhat of a GOP trend there over the past few years. I am more skeptical of how Trump will do with Hispanic voters in the Houston, Dallas, and Austin-SA metros.