Replace usage of strict_exception_groups=False in Nursery.start

[jakkdl]

In preparation for eventual deprecation of strict_exception_groups=False (https://github.com/jakkdl/trio/tree/deprecate_exceptiongroups_false) we probably shouldn't rely on it ourselves in the internal code.

#2611 - the original issue where errors raised before task_status.started() got double-wrapped
#1599 - for why the old nursery can have multiple tasks.
#2844 - the regression error that occured after my initial fix for #2611 in #2826

Fairly straightforward implementation, other than perhaps exactly what to raise when multiple exceptions do occur. The main code that tests this is test_trio_run_strict_before_started, I added the new test after it only to trigger the case where multiple exceptions do pop up in the old nursery.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 99.64751%. Comparing base (7cb15ea) to head (03fee26).

Additional details and impacted files

@@                 Coverage Diff                 @@
##              master       #2938         +/-   ##
===================================================
+ Coverage   99.64717%   99.64751%   +0.00033%     
===================================================
  Files            116         116                 
  Lines          17572       17589         +17     
  Branches        3160        3163          +3     
===================================================
+ Hits           17510       17527         +17     
  Misses            43          43                 
  Partials          19          19                 

Files with missing lines	Coverage Δ
src/trio/_core/_run.py	99.68783% <100.00000%> (+0.00164%)	⬆️
src/trio/_core/_tests/test_run.py	100.00000% <100.00000%> (ø)

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[A5rocks]

I liked how one of the linked PRs proposed warning in .start_soon for that internal nursery (re: todo)

[Zac-HD]

I think making this an internal error immediately is fine - it's possible to do this, but it's never been something we documented or that we think is reasonable to have done for some reason.

[jakkdl]

sounds like we're settling on internalerror, and leaving any eventual warning for #1600 or similar.
removed the comment and expanded the error message to make it clearer for the user that it might be their fault.

[Zac-HD]

Very nice! Merging now 😁