[flang][runtime] Fix runtime CSHIFT of rank>1 array case of negative shift count

[klausler]

Summary

The calculation of the source index was incorrect when a CSHIFT shift
count value is negative, for the implementation of CSHIFT for arrays
with rank >= 2. (The vector CSHIFT is fine.)

flang/runtime/transformational.cpp

void RTNAME(Cshift)(Descriptor &result, const Descriptor &source,
   for (std::size_t n{result.Elements()}; n > 0; n -= dimExtent) {
     SubscriptValue shiftCount{shiftControl.GetShift(resultAt)};
     SubscriptValue sourceAt[maxRank];
     for (int j{0}; j < rank; ++j) {
       sourceAt[j] = sourceLB[j] + resultAt[j] - 1;
     }
     SubscriptValue &sourceDim{sourceAt[dim - 1]};
     sourceDim = dimLB + shiftCount % dimExtent;
-    if (shiftCount < 0) {
+    if (sourceDim < dimLB) {
       sourceDim += dimExtent;
     }
     for (resDim = 1; resDim <= dimExtent; ++resDim) {
       CopyElement(result, resultAt, source, sourceAt, terminator);
       if (++sourceDim == dimLB + dimExtent) {
         sourceDim = dimLB;
       }
     }