Test refactoring part 2

[mjcarroll]

Further cleanups to testing after the subprocess changes.

This does:

• Pushes auxillary test executables down one directory, just to tidy up the integration folder and makes it clear which things are tests vs support executables
• Splits "test utils" (utility code) from "test config" (generated configuration)
• Puts auxillary executable definitions in one location

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (ca954cb) 87.23% compared to head (b9fb754) 87.27%.
Report is 1 commits behind head on gz-transport12.

Additional details and impacted files

@@                Coverage Diff                 @@
##           gz-transport12     #463      +/-   ##
==================================================
+ Coverage           87.23%   87.27%   +0.03%     
==================================================
  Files                  60       60              
  Lines                5211     5211              
==================================================
+ Hits                 4546     4548       +2     
+ Misses                665      663       -2     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[ahcorde]

LGTM, just a small comment about two std includes

[mjcarroll]

There is still a failing homebrew test (that I think is new) and also Windows is failing to clone the repo (@j-rivero?)

[mjcarroll]

@osrf-jenkins retest this please

[mjcarroll]

I found the culprit here, I accidentally switched out the test executables from one that runs indefinitely to one that exits after a few seconds. It seems that on Linux runners, we are fast enough to complete the stress test in the window, where on other platforms we are too slow.

I have switched the test back accordingly.

[mjcarroll]

@osrf-jenkins retest this please

[caguero]

Do you know if the Windows failure is related to this PR?

[mjcarroll]

Do you know if the Windows failure is related to this PR?

I'm not sure. On one hand I haven't seen it before, so maybe my fault, but also seems unrelated to the changes in the PR?

[mjcarroll]

Looks like there are actual multiple failing tests, but they aren't being reported correctly:

The following tests FAILED:
	 47 - INTEGRATION_twoProcsSrvCallStress (Timeout)
	 53 - INTEGRATION_twoProcsSrvCallWithoutInputStress (Timeout)
	 59 - INTEGRATION_twoProcsSrvCallWithoutOutputStress (Timeout)
	 83 - INTEGRATION_playback (Failed)
Errors while running CTest

[mjcarroll]

Looks like Windows failure may have been a flake. I will merge this and continue to monitor.