fix incorrect back-off factor update in dash.py

[zjx20]

No description provided.

[ImAleeexx]

Why should the back off factor be capped at 10 ?

[zjx20]

I should have explained this. Please notice that the back_off_factor variable isn't the actual sleep time, it's a factor.

Let's simply the logic.

back_off_factor = 1
while not closed:
    refresh_wait = 5 # although it could be something else, let's assume that it is 5
    with self.sleeper(refresh_wait * back_off_factor):
        ...
        if not self.reload():
            back_off_factor = max(back_off_factor * 1.3, 10.0)
        else:
            back_off_factor = 1

Suppose self.reload() returns False 5 times in a row, and the values of back_off_factor are 1, 10, 13, 16.9, 21.97, 28.561. The actual sleep times refresh_wait * back_off_factor are 5s, 50s, 65s, 84.5s, 109.85s, 142.805s.
But if we change the max() to min(), the back_off_factor sequence is 1, 1.3, 1.69, 2.197, 2.8561, 3.71293. And the actual sleep time sequence is 5s, 6.5s, 8.45s, 10.985s, 14.2805s, 18.56465.

I believe this is a logical error and min() should be the correct one, because a 10 times longer sleep time for the first failure makes no sense.

Why should the back off factor be capped at 10 ?

Since refresh_wait is the suggested refresh interval, 10 times refresh_wait should be large enough for backing-off.