fix divergence in barriers (#3621)

Without this fix, we see cases in which not all work-items in a thread group end up hitting the same number of barriers, which leads to a hang in OpenCL GPU execution.

platforms/opencl/src/kernels/findInteractingBlocks.cl

@@ -249,6 +249,9 @@ __kernel void findBlocksWithInteractions(real4 periodicBoxSize, real4 invPeriodi
                         neighborsInBuffer -= TILE_SIZE*tilesToStore;
                    }
                 }
+                else {
+                    SYNC_WARPS;
+                }
             }
         }