The modelling makes assumptions about vaccine efficacy against severe disease, disease, and infection for both the AZ & PF vaccines, vaccine uptake rates, and reduction in lockdown measures. As they say "Our results are highly dependent on the assumed (optimistic) vaccine efficacy, uptake, and rollout speed".
There isn't a breakdown of where deaths will occur in terms of the split between those vaccinated and those unvaccinated, or in terms of age profiles. My guess would be that future deaths will still broadly follow the age distribution we've seen up until now, but as you say with additional deaths associated with cohorts who don't get vaccinated for whatever reason and particularly older members of those groups.
The study doesn't suggest that the vaccines are any less effective than has already been shown. It just demonstrates how even slightly imperfect vaccine efficacy has an effect on serious illness across a large population.