I don't know or haven't seen any evidence that polymerase mutations are responsible for the increased R0 of Delta, although it's certainly possible.
What we do know is that the Spike protein is essential for COVID cellular binding.
We also know that spike has mutated between 'vanilla' COVID of 18 months ago and what all vaccines target, and Delta.
This makes sense, because the spike (or S) protein wasn't optimised for binding to Human ACE2 receptors; over time mutations in S (ie with Delta) have lead to increases in binding affinity with ACE2, the result being that Delta is more likely to infect a cell with the same viral load as 'Vanilla'. This would be a perfectly acceptable reason for the uptick in viral loads in Delta and infectivity and hence R0.
We are lucky/the vaccine researchers are smart in that S is quite antigenic (that is, triggers a good immune response, although in a 'natural' infection people will produce antibodies to many parts of the COVID virus), and because the selection pressures driving improvements in ACE2 binding affinity of S have not lead to such a great conformational change that antibody responses are useless (ie antibody-antigen affinity is decreased in Delta, but still effective), our vaccines against the earlier S protein are still effective.
Because of the way evolution generally works, pressures driving S evolution tend to end in a local minima (or maxima, depending on your way of looking at it - i tend to think in terms of conformational entropy so hence Xornot and my use of different terms for the same concept) - or as I mentioned earlier, 'Climbing mount improbable', a term coined by Dawkins in the book of the same name.
It is because it is very hard to get out of these local minima that the human body is full of such oddities, such as an eye that is designed with the nerves in front of the photoreceptors, resulting in a blind spot - it's easier to adapt around the minima to reach a new optimum then it is to reinvent the eye with the nerve in the position that would be logical from a functionality standpoint)
Therefore, it is very very unlikely that the S protein, or the virus, will be able to mutate to a S protein that is so radically different that it escapes antibody-antigen affinity from the vaccines we have today - the S protein would basically have to be completely redesigned, which is really unlikely to occur by the gradual process of evolution. Ie the chances of shaking the RNA coding for S into a new functional protein that works and is a completely new, novel shape, is infinitesimal.
So - basically, total vaccine escape is really really unlikely. Like, number of grains of sand on earth unlikely, if not more so. But polymerase mutations could increase the rate at which the grains of sand are sifted.
It is not something that I personally would lose much sleep over - we don't have any evidence from nature that I am aware of (ie flu virus etc) of highly conserved/selected proteins all of a sudden spitting out a completely new model - basically new diseases that we worry about come from species shifting (ie... let's just say SARS to avoid a debate about COVID origins)