Destroy service clients explicitely

[christophfroehlich]

While trying to fix ros-controls/ros2_control_demos#666 I found out that the cm service_caller adds a client for every call, but never destroys it.
If the spawner is used with a list of controllers, it will create identical service clients in the same node -> might cause issues.

[saikishor]

LGTM

[codecov]

Codecov Report

Attention: Patch coverage is 50.00000% with 1 line in your changes missing coverage. Please review.

Project coverage is 87.71%. Comparing base (7374c43) to head (0045673).
Report is 2 commits behind head on master.

Files with missing lines	Patch %	Lines
.../controller_manager/controller_manager_services.py	50.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1944      +/-   ##
==========================================
- Coverage   87.73%   87.71%   -0.02%     
==========================================
  Files         122      122              
  Lines       13010    13012       +2     
  Branches     1165     1165              
==========================================
  Hits        11414    11414              
- Misses       1165     1166       +1     
- Partials      431      432       +1     

Flag	Coverage Δ
unittests	87.71% <50.00%> (-0.02%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
.../controller_manager/controller_manager_services.py	79.01% <50.00%> (-0.37%)	⬇️

... and 1 file with indirect coverage changes

[bijoua29]

@christophfroehlich Looks good to me. Except I may have a related PR that may obviate the need for this PR. If my PR works, I will submit it within the next day. Either way, I will report here. Do you mind waiting a day before merging this PR? It's OK if you don't want to wait too.

[bmagyar]

May also be the reason of some CI flakiness

[christophfroehlich]

@bijoua29 I can wait with merging this a bit. We made a change to example_13 tests, let's see if this alone will fix the flaky test there.

[bijoua29]

@christophfroehlich I created #1947 which is in conflict with this PR. Please take a look at that PR. I do clean up the service client there in the destructor which is essentially the gist of this PR. I don't know if it helps or hurts with the flaky tests although I don't see how it would affect the tests.

[christophfroehlich]

If there is only a single one per node and service type, then the destruction won't make any difference. But without any change, it happened to have 10 clients and this could be an issue somehow with flaky tests in the ros2_control_demos repo.

[christophfroehlich]

fixed with #1949