Different workaround for false codechecker error

[davidrohr]

@anerokhi : Your workaround turned out to break GPU compilation in some cases, since some types might be only forward-declared in the header, so the implementation of the constructors etc. must be moved to the CXX. I have thus disabled the linter on this line.

[anerokhi]

Hm. How did I forget that I could just disable check for a line :/

[davidrohr]

@Barthelemy : Could you check this compile warning, which currently breaks the CI:

Modules/TOF/include/TOF/TaskDiagnostics.h:45:7: error: 'o2::quality_control_modules::tof::DiagnosticsCounter' has a field 'o2::quality_control_modules::tof::DiagnosticsCounter::mRDHCounter' whose type uses the anonymous namespace

@sy-c : The same for readout with this warning:

src/ReadoutUtils.cxx:165:29: error: control reaches end of non-void function

[davidrohr]

Errors unrelated to this PR, merging.

[Barthelemy]

@davidrohr there is already a PR in QC for that and I have just merged it. AliceO2Group/QualityControl#556

[davidrohr]

@Barthelemy : Thx, can we have a new tag then? Or shall we first test locally whether it works?