So you can do all mathematical operations in binary, but you can't represent all numbers in binary like 0.3, which is a repeating number, and had the same issues as a number like 1/3 in decimal where you can't avoid rounding errors
It's worth noting that 1/3 is also a repeating number in binary. 0.01010101....
I'm not sure what sort of point you think you're making but 0.0100110011 in binary is only 0.065% off from 0.3, but how often would you organically encounter 0.3?
Many fractions in decimal are also repeating numbers or very long trailing numbers, I especially encounter a lot when working with time which is base 60.
That's a matter of convention, not technical definition. A byte can be any number of bits, depending on hardware. For a while 6 bit bytes were common. RFC 791 refers to an 8 bit byte as an octet
French-speaking people do too it seems. On second hand websites in Switzerland you always see that some disks are listed for e.g. 250 Go and others for 250 GB, depending on the first language of the seller.