The lens of Nikon in low light limits the new larger sensors

For a given sensor technology, mostly the noise in pics is independent of the pixels. But for a high density sensor, the higher total noise due to the added pixels of the huge sensor is cancelled out by the higher 'signal', so the S/N ratio is the same. My question: for the best low-light performance, Is it possible to calculate the best possible MP for a set lens. But that won’t be possible, as the legacy wide open lenses cannot determine these new sensors, therefore we end up with the higher S/N of the added pixels, but cannot counterbalance it by a high signal

The MTF of the combined system is found by multiplying the lens MTF at each spatial frequency. Thus a better higher frequency response for lens or camera improves the overall MTF. In the end, it all goes down to when is a lens good enough and the answer is when it gives the image quality that you want. Having large lenses does not do any harm, in fact more pixels are advantage and removes crosstalk issues. The general rule is that more pixels makes a lens better even if the lens itself is poor and will also give higher MTF.

Increasing the camera’s resolution affects the different types of noise as follows:
• decreases the Nyquist noise
• decreases the noise affecting the high frequencies
• increases the noise affecting the low frequencies
Thus, the low light areas will have enhanced frequencies and will positively affect the S?N ratio. You should see Sony's new 24MP APS-C sensor, here the decrease in Nyquist and HF noise is more than the increase in LF noise, and is thus beneficial for a large photographic subjects.

I have a question, if we define noise like that, then a 24mp sensor will have less 'noise' than a 12mp sensor. It will have higher resolution thus. but you forgot to mention about the 'real' noise or the shot/photon noise? If you have same QE, then a 12mp sensor will have lower pixel level shot/photon noise than a 24mp sensor, meaning that the accuracy of the samples have decresed. This is not what we want but maybe you're comparing a downsampled output from a 24mp sensor to the output from a 12mp sensor?

Yes, I am assuming binning or downsampling which is always possible by hardware or software means. if the objective lenses are telecentric; light rays touching the sensor’s surface parpendicularly — there's no thing to prove that a high-resolution sensor would have a worse performance than a low-resolution sensor in capturing purposes.

You are mistaken acc to me. LF noise isn't noise that has a low frequency. It's the noise that affects the areas with low frequency areas, which are the smooth and uniform areas in the object.

The lens has not anything to do with noise. Random variation of individual pixel values is the reason why
more pixels isn't noisier than having fewer. The deviation is totally random than the actual values means that the value of each pixel deviates from the "correct" one by some amount that does not depend on other pixel, not on the resolution of the lens, mostly not on anything.
Due to randomness downscaling an iamge reduces the noise. The averaging of each pixel with others nearby; makes the output closer to correct one, because a pixel that's got positive noise. But this is completely independent of the lens; noise is only depend on the sensor. If you focus your camera badly at an image, it will still give same amt of nice, which can be averaged out by down sampling.