Thanks for all. So R-C plus Schmitt Trigger plus reading until stable would make make it bulletproof, but of course not as fast as Agilent.
Besides: can the probability of Metastability to occur be calculated as function of setup time, sampling rate, slope and frequency of input? Is there a formula somewhere that deals with this? Or a practical example that shows a percentage?

Besides the original request of Stephane REY was to have an adjustment tool for some equipment where as I interprete it, the user would have some feedback of what he is doing, a thing I would solve with a simple approach without thinking twice :)

Werner