PR missing one of the required labels: {'documentation', 'internal', 'dependencies', 'enhancement', 'breaking-change', 'bug', 'new feature'}

All Int8/UInt8/etc. has been added.
The question is now: should we keep support for int and float? In this case, it must be clearly documented than int would be equivalent to Int64 (arbitrary choice), and float to Float64 (arbitrary choice).
Otherwise, the use can add the support with only 4 lines of Python:

serializer(lambda i: ZBytes.serialize(Int64(i)), target=int)
deserializer(lambda zbytes: zbytes.deserialize(Int64), target=int)
serializer(lambda f: ZBytes.serialize(Float64(f)), target=float)
deserializer(lambda zbytes: zbytes.deserialize(Float64), target=float)

I'd go for keeping int and float and mapping it to Int64 and Float64, which is kind of the same behaviour of usize on 64bit architecture on Rust.
It needs to be clearly documented.

I've added ZBytes documentation. z_bytes.py example is in the PR following.