bpo-46993: Speed up bytearray creation from list and tuple

[kumaraditya303]

Speeds up bytearray creation from list and tuples.

Benchmark:

importpyperfrunner=pyperf.Runner()
runner.timeit(name="bench bytearray",
stmt="bytearray([42]*10000)",)

Result:

Mean +- std dev: [base] 74.8 us +- 5.5 us -> [patch] 53.2 us +- 3.3 us: 1.41x faster

https://bugs.python.org/issue46993

[gvanrossum]

Yes, much better.

I suppose there's a way to observe the new behavior, if the jump to the slow path must be taken: the offending subtype-of-long-instance will see that its __index__ method is called twice. But I see no reason this would matter.

[kumaraditya303]

Thanks @asvetlov and @gvanrossum