Update CI config

[JStech]

• Use industrial_ci UNDERLAY again, after learning that we did Use UNDERLAY feature of industrial_ci #76 wrong.
• Unify config with the one from main MoveIt repo

[codecov]

Codecov Report

Merging #83 (f8e0547) into master (034f951) will increase coverage by 6.42%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master      #83      +/-   ##
==========================================
+ Coverage   72.75%   79.16%   +6.42%     
==========================================
  Files          10        9       -1     
  Lines         554      499      -55     
==========================================
- Hits          403      395       -8     
+ Misses        151      104      -47     

Impacted Files	Coverage Δ
...t/handeye_calibration_solver/handeye_solver_base.h
...ye_calibration_target/src/handeye_target_aruco.cpp	87.78% <0.00%> (+4.45%)	⬆️
..._calibration_target/src/handeye_target_charuco.cpp	88.78% <0.00%> (+5.76%)	⬆️
..._calibration_solver/src/handeye_solver_default.cpp	64.62% <0.00%> (+6.89%)	⬆️
...t/handeye_calibration_target/handeye_target_base.h	88.08% <0.00%> (+7.59%)	⬆️

---

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 034f951...f8e0547. Read the comment docs.

[rhaschke]

I took the chance to unify the CI config with that of the MoveIt main repo.
I also added a Noetic build. However, this fails. Please have a look!

[tylerjw]

While updating this, do we want to add the pre-release test here too?

[JStech]

I also added a Noetic build. However, this fails. Please have a look!

That test failure is because OpenCV 3.2, the default version for Ubuntu 18.04, has a buggy ArUco board pose detector. See discussion here, #5. I added a check to use the incorrect pose for 3.2 and the correct pose otherwise.

[rhaschke]

From my point of view, this is ready to merge.
I'm not sure though, that we should consider wrong results for the unit tests against OpenCV 3.2. Alternatively, we could skip the test completely (in this case). However, then the remaining code isn't tested as well. Hence, I'm fine with the current solution as well.