Better _binary module use

[homm]

• remove extra i8 calls where input is proved bytes[] or int. Most of i8 calls are not required in py3.
• use offset for all binary input functions instead of slicing which should be slightly faster especially when slicing "until the end" is used (s[4:]).

[homm]

In [1]: [c for c in b'test']
Out[1]: [116, 101, 115, 116]

In [2]: list(b'test')
Out[2]: [116, 101, 115, 116]

[homm]

I don't have enough permissions to restart build on Travis (

[hugovk]

I don't have enough permissions to restart build on Travis (

Maybe you do: try logging out and back in to Travis, that brought back the restart buttons for me recently.

[homm]

Thanks, it works)

[radarhere]

Is it just me, or is byte = byte[0] not as clear as byte = i8(byte) about what is happening here?

[homm]

You are right that i8() looks more clear at first glance. Mostly because indexing had changed from Py2 to Py3 for bytes. However, this saves one function call and type checking on each value. Also, IDEs handle types correctly and show variables as int.

[homm]

As an alternative, it is possible to change i8 and use it everywhere for consistency:

def i8(c, o=0):
    return c[o]

[radarhere]

I've created a rebased version of this as #5156