I believe that in the image assembly pipelines for processing these astronomical images, they already do take into account / use the "dithering" patterns that you're hinting at. (Often the telescope will be pointed in a pattern with sub-pixel offsets over multiple exposures to do exactly this).
However, 2 factors:
1) there is an intrinsic limit I believe to how much more resolution you can recover (maybe a factor of approx. 2x?), for a lot more exposure time needed However, also at these faint levels of brightness you're also competing against intrinsic photon and sensor noise)
2) practically, given the value of the telescope's time and not much more to be gained (science-wise) from achieving this next order of spatial resolution, they want to spend the time on other new targets instead of sitting on the same patch for much more time.
However, 2 factors:
1) there is an intrinsic limit I believe to how much more resolution you can recover (maybe a factor of approx. 2x?), for a lot more exposure time needed However, also at these faint levels of brightness you're also competing against intrinsic photon and sensor noise)
2) practically, given the value of the telescope's time and not much more to be gained (science-wise) from achieving this next order of spatial resolution, they want to spend the time on other new targets instead of sitting on the same patch for much more time.
(you can even try this at home: https://petapixel.com/2015/02/21/a-practical-guide-to-creati... )