fix the bug that the fill level of the interface cell can fall between 0 and eps.

src/lbm/free_surface/dynamics/StreamReconstructAdvectSweep.h

@@ -151,6 +151,8 @@ void StreamReconstructAdvectSweep< LM, FlagInfo_T, abortOnHighFillLevelChange >:
    {
       for ( auto y = xyz.yMin(); y <= xyz.yMax(); ++y )
       {
+         //TODO the isLeft flag should be true for each block at the beginning, so it might be that the mass is inherently increased.
+         //TODO the top and bottom interface are not equally distributed as at the start, the mass is gathering around 0.9 -->//FIXME
          real_t fillLevelDragger = fillSrcField->get(xyz.xMin()-1,y,z);
          bool isLeft = true;
          for ( auto x = xyz.xMin(); x <= xyz.xMax(); ++x )
@@ -169,6 +171,9 @@ void StreamReconstructAdvectSweep< LM, FlagInfo_T, abortOnHighFillLevelChange >:
                   {
                      fillDstField->get(x,y,z) = fillSrcField->get(x,y,z) - amount;
 //                     if ( flags_.isInterface(flagField->get(x,y,z)) && y == cell_idx_c(19) && z == cell_idx_c(16) ) {WALBERLA_LOG_DEVEL("reduced FL = (" << x << "," << y << "," << z << ")" );}
+                     // Here it can be possible that the fill level is greater than eps but smaller than 0
+                     // ==> Thus, it has to be ensured that the cell is going to be converted to a gas cell
+                     if ( fillDstField->get(x,y,z) <= real_c(0) ) {fillDstField->get(x,y,z) = real_c(-0.1);}
                   }
                   else
                   {