Threadpool improvements

[wirew0rm]

Backport of changes from graph prototype. See original issue:
fair-acc/gnuradio4#67

Some noteworthy changes:

• emscripten compatibility: std::jthread -> std::thread, ranges usages
• fix race condition on thread shutdown
• fix [0pts] Fix starting and stopping threads in ThreadPool #243 (starting and stopping threads in ThreadPool)

Also enables execution of emscripten unittests via ctest and unbreak failing emscripten tests.

[RalphSteinhagen]

Linked to fair-acc/gnuradio4#67

Thanks for enabling the WASM test work and following up on the ThreadPool adjustments. Nice!

[codecov]

Codecov Report

Patch coverage: 47.91% and project coverage change: +0.40 🎉

Comparison is base (418003b) 55.29% compared to head (3fb355a) 55.70%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #298      +/-   ##
==========================================
+ Coverage   55.29%   55.70%   +0.40%     
==========================================
  Files          68       65       -3     
  Lines        7216     7174      -42     
  Branches     2667     2613      -54     
==========================================
+ Hits         3990     3996       +6     
- Misses       1526     1527       +1     
+ Partials     1700     1651      -49     

Impacted Files	Coverage Δ
src/client/include/ClientContext.hpp	55.81% <ø> (ø)
src/client/include/RestClientNative.hpp	45.41% <ø> (-0.53%)	⬇️
src/majordomo/include/majordomo/Settings.hpp	100.00% <ø> (ø)
src/core/include/Debug.hpp	43.58% <7.69%> (ø)
src/majordomo/include/majordomo/RestBackend.hpp	17.53% <8.10%> (-0.02%)	⬇️
src/core/include/ThreadAffinity.hpp	59.57% <14.28%> (ø)
src/majordomo/include/majordomo/Broker.hpp	53.10% <34.50%> (+5.74%)	⬆️
src/client/include/Client.hpp	42.42% <41.66%> (+4.50%)	⬆️
src/client/include/MockServer.hpp	58.44% <46.15%> (+1.68%)	⬆️
src/majordomo/include/majordomo/MockClient.hpp	51.51% <46.15%> (+8.43%)	⬆️
... and 5 more

... and 1 file with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

[wirew0rm]

Added -fprofile-update=atomic to the coverage compiler options, see documentation: https://gcc.gnu.org/onlinedocs/gcc-9.3.0/gcc/Instrumentation-Options.html, to prevent the problem of negative coverage line hits due to concurrent coverage counter updates as described in the issue linked from the error message:

 gcovr.gcov_parser.NegativeHits: Got negative hit value in gcov line 'branch  1 taken -1' caused by a bug in gcov tool, see https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68080. Use option --gcov-ignore-parse-errors with a value of negative_hits.warn, or negative_hits.warn_one.

https://github.com/fair-acc/opencmw-cpp/actions/runs/5142318355/jobs/9257131141#step:14:1360

Since this is not deterministic, I'm not completely sure that this will fix the issue. According to the documentation it gcc should automatically select atomic if -lpthread is used, but is not completely clear about whether this applies generally or only for -fprofile-update=maybe-atomic. If it reoccurs in the future the other possible problem would be unjoined threads which break coverage.