You are saying the magic word: prototypes. That explains the huge budgets, time and cost overruns. They certainly have an important place in science, because sciences often is about doing things the first time. But there is also a place for production runs. Because you are no longer just making prototypes but start to reuse the experiences and expenses.
Hubble is 30 years old, and there isn't any direct replacement in sight at all. At latest now, with cheap launches and much improved technology, it is more than time to replace it and as it is more or less a direct replacement, use the scale of production. Don't build one telescope for 3 billion, build 10 at like 100-200 million each. If one ore more of them fail, so what?
Well the reason that we would build the one 3 bn telescope, rather than 10 100-200 m telescopes, is because they can tell you fundamentally different things about the Universe.
In many ways, to move forward, we must go bigger with the tech, not wider. What you are essentially suggesting is a survey - we already have instruments and telescopes to do exactly that, with new ones coming online soon (see the LSST). Deep observations like Hubble was meant to do can actually be done better from the ground now (e.g. the VLT, ELT, etc.). As such, focus has shifted to those for optical observations.
The community at large, however, has decided where it would like to spend its budget. That was on Webb for the IR wavebands, to study so-called 'high redshift' objects that Hubble struggles to see, and leave the more detailed imaging to those ground based observatories.
Look, I would love another Hubble. That would be really useful! And that's why WFIRST exists. But an in-place replacement doesn't make sense. The tech on Hubble (not necessarily the mirror, but the _instruments_, the spectrographs, the interferometers) is really old. You can't take an off-the-shelf component, because it needs to be built to withstand space. So you need to create a one-off, 'prototype', space-hardened instrument. It is also not a case of worrying about them 'failing', but you also must worry about them slowly drifting out of a calibrated range. On the ground, you could fix the instrument, or re-calibrate it. That's not possible in space. That's not a 'random' style error, where we can get out of it by upping the number of them - they are systematic, so all 10 of your proposed telescopes would be ruined.
Hubble is 30 years old, and there isn't any direct replacement in sight at all. At latest now, with cheap launches and much improved technology, it is more than time to replace it and as it is more or less a direct replacement, use the scale of production. Don't build one telescope for 3 billion, build 10 at like 100-200 million each. If one ore more of them fail, so what?