In terms of infrastructure gauging and the like, I suspect that our railways are most akin to those of Japan, where the only real increases in structure gauge is through the construction of new main lines (shinkansen). A few japanese traditional routes do use double-decker cars, but only for first class, where seating capacity is important, and there is no requirement to make arrangements for standing passengers. Where standing capacity is important (i.e., the rest of the train), the japanese still use single-level cars with better headroom, that can support hanging straps and grab rails.
Again, similar to the situation in Japan, most of the routes that could benefit from the capacity of larger stock have one other issue that severely limits throughput (and hence overall line capacity), and that is dwell time. We already have an issue in that our stock has to balance out the following:
- Standing capacity
- Seated capacity
- Maximising door openings (to minimise dwell time)
- Providing an easy route between the seated area and the vestibules
- Locating the doors away from the largest gaps on curved platforms (of which we have far more than most)
- Bogie density (and hence weight of the train and track forces)
- Increased complexity due to a higher door density (more maintenance and failures due to door problems)
Given all these factors, it is clear that we would not get double deck stock that can provide a sufficiently fast access/egress time, and then we'd still have to take into account infrastructure works.
To give an example of how dwell time is important, a 14 second extension of dwell time on a network like Crossrail or Thameslink would reduce the capacity of the tunnel by 2 trains per hour, or 1800 seats per hour and direction.