absolutly pandas conform tested - forward also works like pandas forward-hack, center is also tested

fixed constants detection. -there was the famous `ramp-up` error, because of the laggy implementation of pd.rolling ;)

bugfix: unmasking reinjected the new (i.e. nan) instead of the old values (this should have been caught by a test)

function, as dynamically assigned variables are missing otherwise

        failed, as they were evaluated during 'compile' time instead
        of execution time