Only suggest adding a zero to integers with a preceding dot when the change will result in a valid floating point literal. For example, `.0x0` should not be turned into `0.0x0`.
