Good observations and good points... all !
Where I worked (B&L, Kodak, ITT, Harris, Excelis, Navy... others) we made mirrors from rather small to several meters. The biggest difficulty (and therefore time, cost) driver, as a function of SIZE... is the wavelength-normalized accuracy needed on the optical surface of the mirror.
My explanation goes like this >>>
Consider a nice, ten-inch, smooth, 1/10-wave paraboloidal Primary Mirror. Use it to view the heavens, and everything is perfection... as good as the laws of physics permit. It is, functionally, a ~perfect~ mirror.
Now, just take that exact same mirror, and grow it ten times bigger. It is now a sorely-deficient, hundred-inch, rough, 1-wave paraboloidalish Primary Mirror. Use it to view the heavens, and everything is blurry, aberrated, hairy... as bad as the terribly-flawed Hubble mirror produced at its disastrous first-light.
So, in that sense, a mirror that is ten times as big has to be ten times as accurate... just to perform as well as the smaller one. Think of a good 100-incher as comprising one hundred ten inchers, all connected together, and needing to deliver simultaneously great wavefront, plus tip-tilted, pistoned and radiused to match one common focal point. That's asking a LOT. Add to those considerations, the amplified difficulties of mounting, aligning, pointing... the whole task is extremely non-linear. More like cubic, quartic, quintic?! All of this because the wavelength of light does not scale up with aperture! So, our perfection yardstick... remains fixed! Tom