Add data size units as first-class citizens about units

[JaapAap]

As these are discrete units, would this allow them to be integer-based?

[nholthaus]

I don't think so, not in the general case.

Imagine you did:

gigabyte_t size = 256_kB;

You'd be expecting to get 0.000256, not 0. Or,

gigabyte_t size_GB(1.2);
byte_t size_B = size;

you'd be expecting size == 1288490189;

[nholthaus]

Right now there are basically two options for integral units:

1. Redefine the underlying type globally
2. Leave is as an exercise to the user to implement their own.

I'm starting to think that a third, more user-friendly option is needed, but I'm not sure what that should look like.

[JaapAap]

Understood. I currently use similar units that are rather defined in a similar way to how 'chrono' implements time units - I can then use them to e.g. set sizes of buffers (rather than involving them in arithmetic), which is a different use case I suppose.

[nholthaus]

It seems appropriate to implement this using the ISO/IEC 80000 standard. I.e.

1_MB; - 1000² bytes
1_MiB; - 1024² bytes

Any qualms?

[JaapAap]

None whatsoever- like it.