#37 Fix bug in "is flip needed" test (isFlipNeeded)

Previously flip was rejected only if both vertices of flip-candidate belong to the super-tri.
The change rejects the flip if at least one vertex belongs to super-tri and original edge does not touch super-tri.
If both original edge and flipped edge touch super-tri: use normal circumcircle test as a tie-breaker.

Add two test files for regression testing.

CDT/include/CDT.hpp

@@ -451,16 +451,23 @@ bool Triangulation<T>::isFlipNeeded(
     const Triangle& tOpo = triangles[iTopo];
     const Index i = opposedVertexInd(tOpo, iT);
     const VertInd iVopo = tOpo.vertices[i];
-    if(m_superGeomType == SuperGeometryType::SuperTriangle)
-        if(iVert < 3 && iVopo < 3) // opposed vertices belong to super-triangle
-            return false;          // no flip is needed
     const VertInd iVcw = tOpo.vertices[cw(i)];
     const VertInd iVccw = tOpo.vertices[ccw(i)];
     const V2d<T>& v1 = vertices[iVcw].pos;
     const V2d<T>& v2 = vertices[iVopo].pos;
     const V2d<T>& v3 = vertices[iVccw].pos;
     if(m_superGeomType == SuperGeometryType::SuperTriangle)
     {
+        if(iVert < 3 || iVopo < 3) // flip-candidate edge touches super-triangle
+        {
+            if(iVcw < 3 || iVccw < 3) // but so does original edge
+            {
+                // let the normal circumcircle test decide
+                return isInCircumcircle(pos, v1, v2, v3);
+            }
+            return false; // no flip is needed
+        }
+
         if(iVcw < 3)
             return locatePointLine(v1, v2, v3) == locatePointLine(pos, v2, v3);
         if(iVccw < 3)

visualizer/data/regression_issue_38_wrong_hull.txt

@@ -0,0 +1,37 @@
+36 0
+0.15147567518991 -5.144230291488
+0.10442 -5.06098
+0.1228735248885 -5.12897066233
+0.1322969035965 -5.141286788425
+0.115882234089 -5.115648852375
+0.13262891777369 -5.119598039304
+0.11311813200326 -5.138343285364
+0.111139345548 -5.09040217055
+0.121648346645 -5.09478239621
+0.1152696449675 -5.098554719315
+0.1303134549875 -5.106097900345
+0.10889094329024 -5.1023270424231
+0.142052296482 -5.131914165395
+0.154751587595 -5.127544895745
+0.1484019420385 -5.129729530575
+0.1389785633305 -5.117413404475
+0.1563895437975 -5.119202197875
+0.150039898241 -5.1213868327
+0.140515429906 -5.12466378494
+0.145328208887 -5.11522876965
+0.1516778544435 -5.113044134825
+0.1580275 -5.1108595
+0.107779672774 -5.075691085275
+0.112085873903 -5.07104664934
+0.1514750401785 -5.061655309135
+0.13440575 -5.08723775
+0.110784 -5.063616
+0.11338774780615 -5.078477298681
+0.1682004187975 -5.131013072875
+0.18164925 -5.13448125
+0.1921660803575 -5.05969461827
+0.1750967901785 -5.085277059135
+0.221046245536 -5.045923052405
+0.31820357142857 -4.9298217230895
+0.2499264107145 -5.032151486545
+0.284064991072 -4.980986604815

visualizer/data/regression_issue_38_wrong_hull_small.txt

@@ -0,0 +1,7 @@
+6 0
+0.15147567518991 -5.144230291488
+0.10442 -5.06098
+0.1228735248885 -5.12897066233
+0.1322969035965 -5.141286788425
+0.115882234089 -5.115648852375
+0.13262891777369 -5.119598039304