You can almost think of this as something similar to vignetting in the final data products. Certain areas will have lower statistics and especially lower temporal resolution based on the season depending on where they are relative to the horizon near twilight.
So, a single image could be lost, but there is supposed to be 1000 or so good images of that area over the survey, about 100 a year. With the satellites, potentially you’ve now lost 3–8 images a year for any given section of the sky (probably more near the equatorial plane), lowering your statistics of the entire survey 1-10%, depending on the declination.
I’m spitballing numbers, there are actual papers you could read though.
Rubin is “wider, deeper, faster”. This reduces all of those dimensions to some extent, but especially the deeper.
Galaxies are tiny little things in a telescope's view and a satellite always in that view, even if it's only one out of so many shots, could make it harder to glean useful data from an interaction between two galaxies like that. That beam is moving at near the speed of light and a telescope like this could tell us if there are fleeting but significant changes in all that power interacting with matter. Unless a satellite interferes with that light at just the wrong time.
The article says Rubin pictures are 30 seconds exposure, which should leave at least 29* seconds of data.
Of course, there will unavoidably be some degradation.
* This number is backed solely by my personal intuition