if you refer to the suffixes 99999999, 00000004 etc.
These are numerical rounding-errors produced by the floating point operations and there is no easy way to avoid them, except if we use the arbitrary precision library with a penalty in overall performance.
Apart from being visually not pleasant they do should not create any error in the calculations since this is how the CPU will see the numbers.
Another approach to get rid of them, is to reduce the precision in flair and python will round them to the closest round decimal number, however doing so we had troubles in our LHC accelerator lines when small rotations (mrads) where used for the magnets
In the development version of flair that will be released soon, it has some extra checks to avoid most of those situations but there will be always a corner case that will produce it