Show how context managers do not support suspended execution

gen_context.py

class CM:
    mode = 0
    
    def __init__(self, m):
        self.m = m

    def __enter__(self):
        self.old = CM.mode
        CM.mode = self.m

    def __exit__(self, *_):
        CM.mode = self.old

def fizz():
    with CM('fizz'):
        for _ in range(3):
            yield f'fizz({CM.mode})'

def buzz():
    with CM('buzz'):
        for _ in range(3):
            yield f'buzz({CM.mode})'

for f, b in zip(fizz(), buzz()):
    print(f, b)

Note how fizz iterations have the wrong mode after the first iteration of buzz.

As in the await code, one would expect the body of each generator to use the same context, but they do not:

fizz(fizz) buzz(buzz)
fizz(buzz) buzz(buzz)
fizz(buzz) buzz(buzz)

The first fizz iteration is the only one with the correct mode - fizz(fizz).

---

The expected output is:

fizz(fizz) buzz(buzz)
fizz(fizz) buzz(buzz)
fizz(fizz) buzz(buzz)

Where each iteration of both generators occurs in the correct context.