As I said, the problem for eric is that of quantification, including knowledge of the degree of flicker that is required to result in problems for his daughter.
No credible amount of smoothing will get anywhere near eliminating 50 Hz fluctuations in the light intensity - whether detected by 'ad hoc' methods you mention or more scientifically (like the sensor+'scope method I mentioned) - so the issue is not to determine whether or not flicker is present (which it always will be, if one looks hard enough), but whether it is of a magnitude to represent a problem.
Kind Regards, John