Cancel a cancel scope upon entry if its deadline is in the past

[oremanj]

Fixes #320. Now that #901 is implemented, the concerns discussed in #835 don't apply.

[codecov]

Codecov Report

Merging #909 into master will increase coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff            @@
##           master    #909      +/-   ##
=========================================
+ Coverage    99.4%   99.4%   +<.01%     
=========================================
  Files         102     102              
  Lines       12679   12681       +2     
  Branches      926     927       +1     
=========================================
+ Hits        12603   12605       +2     
  Misses         56      56              
  Partials       20      20

Impacted Files	Coverage Δ
trio/_core/tests/test_run.py	100% <ø> (ø)	⬆️
trio/_core/tests/tutil.py	100% <ø> (ø)	⬆️
trio/_core/_run.py	99.71% <100%> (ø)	⬆️

[njsmith]

Test failure is an intermittent failure in test_open_tcp_listeners_backlog:

• I made a note in Tracking issue: intermittent test failures #200 (comment)
• I clicked the button on azure to "retain" the build, so we'll have the logs for reference
• I edited the top comment on Tracking issue: intermittent test failures #200 to note the requirement to "retain" the build
• Clicked "Queue" to queue a re-build (I think?)

[njsmith]

OK, so the "Queue" button is a trap. It definitely didn't do what we want.

Now I tried clicking the "Rebuild" button that's hidden behind the "..." menu. Maybe that will work better?

[oremanj]

It looks like the rerun worked (https://dev.azure.com/python-trio/trio/_build/results?buildId=152) but github didn't notice, so I guess I'll do the close-and-reopen dance.

[njsmith]

Hmm, azure's UX could definitely use some improvement here :-)

Anyway, the PR seems fine :-). I'm not sure it makes a huge difference either way, but it's certainly a bit closer to what people expect, so why not.