Fix the OpenCL portion to match the c++ code.

Fix an undiscovered bug in the c++ code.

modules/imgproc/src/clahe.cpp

@@ -215,7 +215,7 @@ namespace
                 if (residual != 0)
                 {
                     int residualStep = MAX(histSize / residual, 1);
-                    for (int i = 0; i < histSize; i += residualStep)
+                    for (int i = 0; i < histSize && residual > 0; i += residualStep, residual--)
                         tileHist[i]++;
                 }
             }

modules/imgproc/src/opencl/clahe.cl

@@ -201,7 +201,10 @@ __kernel void calcLut(__global __const uchar * src, const int srcStep,
         tHistVal += redistBatch;
 
         int residual = totalClipped - redistBatch * 256;
-        if (tid < residual)
+        int rStep = 256 / residual;
+        if (rStep < 1)
+            rStep = 1;
+        if (tid%rStep == 0 && (tid/rStep)<residual)
             ++tHistVal;
     }