> It's not clear to me if anyone benchmarked to see if the
> conversion to a macro had any measurable performance benefit.

I tested on that day, also use this command: 

python.exe -m pyperf timeit -s "from collections import deque; consume = deque(maxlen=0).extend; r = range(256)" "consume(r)"  --duplicate=1000

I remember the results are:
inline function: 1.6  us
macro version  : 1.27 us
(32-bit release build by MSVC 2017)

Since the difference is too obvious, I tested it only once for each version.