[RFC] multicall: hookwrapper: use yielded results, support firstresult=True

[blueyed]

Ref: pytest-dev/pytest#5301 (comment)

Currently results from a hookwrapper are ignored, this changes the behavior (see test_hookwrapper_order).
Should this be explicitly enabled, e.g. via a new "use_hookwrapper_result" kwarg for the hookwrapper? As the existing test shows "yield" might be used with a result, which currently is ignored.

A hookwrapper currently can set the return value / outcome via force_result, but it would still run non-hookwrappers before.

[codecov-io]

Codecov Report

Merging #246 into master will decrease coverage by 0.7%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master     #246      +/-   ##
==========================================
- Coverage   92.91%   92.21%   -0.71%     
==========================================
  Files          14       14              
  Lines        1651     1541     -110     
  Branches      115      116       +1     
==========================================
- Hits         1534     1421     -113     
- Misses        101      102       +1     
- Partials       16       18       +2

Impacted Files	Coverage Δ
src/pluggy/callers.py	79.36% <100%> (+0.16%)	⬆️
testing/test_multicall.py	99.23% <100%> (-0.05%)	⬇️
testing/test_hookcaller.py	77.39% <0%> (-4.81%)	⬇️
src/pluggy/manager.py	92.3% <0%> (-2.09%)	⬇️
testing/conftest.py	91.66% <0%> (-1.67%)	⬇️
testing/test_deprecations.py	90.62% <0%> (-0.56%)	⬇️
testing/test_helpers.py	81.48% <0%> (-0.34%)	⬇️
testing/test_pluginmanager.py	95.67% <0%> (-0.33%)	⬇️
testing/test_invocations.py	98.34% <0%> (-0.27%)	⬇️
... and 3 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 1a341b6...76838b1. Read the comment docs.

[RonnyPfannschmidt]

this should get a new name as its a new flow control

hookdecorator sounds possibly sensible - we should also figure how they order relative to regular hook wrappers (as its potentially critical to be either before them or always within them)

[blueyed]

Yeah, ok. Not planning to do this anytime soon - closing.

I've used two subplugins with pytest-dev/pytest#6448 now.