Publicly released Presidential preference polls don’t have a great track record. Many of them gave Hillary a lock on the Electoral College and badly mispredicted by forecasting 5–10 point Hillary wins in Wisconsin, Michigan,and Pennsylvania. They told her there was no need to go to Wisconsin and no need to boost advertising in Michigan. She and the self-deluded morons running her campaign and all her media acolytes kept patting themselves on the back before the election on the basis of those polls.
Democrats should take no great solace in the latest polls until the pollsters explain what changes they have made in their process. Even then, there is no way to validate their fixes until the next Presidential election. Your opinion hit on a key problem: polls based on landline responses are not getting an unbiased representative sample. Another issue: working people are out working during the day and may be taking care of kids or going out or watching TV at night. So who exactly is answering these calls and answering the pollsters? Possibly a very atypical group, possibly over-weighted with Left wing sympathizers. Then, of the people who do answer, how many don’t want to look uncool by admitting some sympathy for Trump? Overall much more needs to be done to improve the reliability of polling. Some particular polls may have done the work. But we hear little about what has been done to make the polls more accurate. Further the media seldom reports the error bars and even less often how those error bars are derived. Instead it is focused on broadcasting news of favorable poll results and prematurely declaring victory again just like it did in 2016. Of course the Democrats want it to be true, but are they stupid enough to believe their own propaganda? Yes, they did so in 2016 and some are doing it again.