1.) The given link contains only excerpts of a lecture.
2.) There is a formula to calculate the probability of metastability but it deals only with FF directly connected.
A delay should increase the probability of M. and a gain should decrease it. But how much???
for essentially delay insensitive bibliography; however, some references are general. They might include both.
Incidentally, the web site in my previous posting gives a hint how to accomplish the delays, namely, over a counter of the clock pulses.
Also, visit