“using CMOS we collect in 1h the same amount of photons, like with CCD in 20h” – this is obvious false. Sensor’s quantum efficiency tells us how many photons we will have, and that is the same level for CMOS and CCD – that differs of course from one sensor type to another, but not as much. Low read noise for CMOS tells us, that it will handle short exposure times better. But when exposure gets longer the difference between CMOS and CCD lowers.
“the 1 electron read noise stated in CMOS specifications is marketing babble plus it decreases with gain increase” – yes, it is a little bit against logic – we raise gain, and read noise is lower. Well, the unit matters here – that is mapped to signal electrons. When we raise gain, the read noise in the frame (expressed in ADU) raises as well, but after dividing it by camera gain it turns out that read noise is effectively lower, because large part of read noise is created during or after signal amplification, thus raising gain will affect signal more, than read noise . The same effect we have in consumer DSLRs.
“12 bit analog digital converters are shoddy” – yeah, that is separate article for this Is dozen of bits good enough?
“I used to have perfect optics…” – this is my favorite so far, and is related to usually small pixel size of CMOS sensors. Out of nowhere it turns out, that my so far decent optics, that produced perfect images with 10um pixel size camera, works much worse with 3um pixel size camera. Huh, really? Does camera increase field curvature? Or enlarges aberrations? That would be some nice trick! Fortunately it does not. Small pixel allows us to reveal more real details in the picture, but also, in the same time more optics defects is visible. But it does not mean, that this flaws have not existed before, or they have been smaller, not at all. They were just undersampled with large pixel, therefore not or barely visible. With small pixel you will se more, either good (if optics allows) or bad things. Small pixels do not judge.