PR for issue #719

[MartiDvrk]

• Call FlushSquaredValue() explicitly at the beginning of EndOfEventAction(). In fact, there is an implicit flushing implemented in ScoreSquaredValue() triggered in SteppingAction that flushes deposit in a voxel whenever a deposit by a new event is made. Consequently, the deposit made by the last event in any voxel remains unflushed and we must trigger the flushing explicitly.
• At the same time, once the value is flushed, we must make sure that the voxel in the thread local image is reset to zero. So at the end of FlushSquaredValue(), we must iterate over the image and reset all pixels to zero.

[dsarrut]

not sure : the flush is implicit each event and should NOT being explicit for all the pixels at each event, otherwise it will be too long. The flush should be only 1) at the end of run or when you compute the uncertainty, within the "if NbOfEvent >= NbEventsNextCheck" part.
Indeed, the 1D data must be set to zero within FlushSquaredValue () event after each Run

[MartiDvrk]

Modified to Flush just before uncertainty is calculated.

[dsarrut]

seems good ! An additional change please (the last ;) !). Can you set to zero all the values of squared_worker_flatimg in FlushSquaredValue, after the end of the loop ? I think when there are several runs or when the uncertainty is computed, it must be set to zero to avoid "flushing" two times the values in a subsequent event. Thanks !
PS: dont bother about the failing test "test081_simulation_optigan_with_random_seed" it is a knowing issue, under investigation.