The other problem with these is the big super high quality optical mirrors. Going to a production line with those would be really interesting! I don't think you'd get a big enough economy of scale to mass produce them in a just-in-time pipeline though.

Once you've designed and built the grinding machine (I saw a picture of it, it fills a special room) you've done most of it. Same for the custom test rig.

SpaceX shows what happens to costs when you re-use designs and tooling.

True but there's also decreasing marginal utility for each duplicate. A second telescope in any given class would be awesome but it's always going to be less groundbreaking and useful than the first one.

The universe is so big, and so many things to look at, I find it hard to believe we couldn't productively use 100 Hubbles.

100 Hubble's would be great but 100 Hubble's and nothing else would not be

Well if we had built a warehouse full of Hubbles, then launching a duplicate would make a lot of sense, but we didn't and now the cost of making a replacement after all the custom tooling and talent has been lost is just as expensive as the first time around.