When trying to port humansize code to bytesize, I noticed that if the purpose is only for formatting to byte size, many inputs are usize or even positive isize. Forcing each place to case to u64 can be wordy.
Not sure if it can be possible to have a lossy constructor, or I should use humansize in this case.
Previous discussion - #7
humansize trait - https://docs.rs/humansize/2.1.3/humansize/struct.ISizeFormatter.html & https://docs.rs/humansize/2.1.3/humansize/trait.ToF64.html