losslessly convert int64/uint64 represented as high/low part to/from string (only decimal) using double precision floating point arithmetics