#14698 Guard against NaN/Inf in ensemble parameter correlation

Realizations whose calculated summary vector value is NaN or Inf at
the selected timestep are now excluded from the Pearson correlation
computation instead of propagating NaN through the whole result,
which corrupted the correlation tornado plot. Also guard the final
correlation value against NaN, since the previous check only compared
against +Infinity.

RigEnsembleParameter::stdDeviation() sums squared parameter values,
which can overflow to +Inf for extreme (but finite) magnitudes,
turning normalizedStdDeviation() into NaN. Since NaN compares false
against the variation epsilon, such parameters were silently treated
as having no variation and dropped from parameterCorrelations(),
producing an empty tornado plot for that parameter and calculated
vector combination. Fall back to a range-based variation estimate
when the standard deviation calculation overflows, so the parameter
is retained and the Pearson correlation guard above can take over
instead.

ApplicationLibCode/ProjectDataModel/Summary/RimSummaryEnsemble.cpp

@@ -569,7 +569,7 @@ std::vector<std::pair<RigEnsembleParameter, double>>
                 }
             }
         }
-        if ( closestValue != std::numeric_limits<double>::infinity() )
+        if ( std::isfinite( closestValue ) )
         {
             caseValuesAtTimestep.push_back( closestValue );
 
@@ -591,7 +591,7 @@ std::vector<std::pair<RigEnsembleParameter, double>>
     {
         double correlation = 0.0;
         double pearson     = RigStatisticsTools::pearsonCorrelation( parameterValuesPair.second, caseValuesAtTimestep );
-        if ( pearson != std::numeric_limits<double>::infinity() ) correlation = pearson;
+        if ( std::isfinite( pearson ) ) correlation = pearson;
         correlationResults.push_back( std::make_pair( parameterValuesPair.first, correlation ) );
     }
     return correlationResults;

ApplicationLibCode/ReservoirDataModel/RigEnsembleParameter.cpp

@@ -58,7 +58,16 @@ double RigEnsembleParameter::normalizedStdDeviation() const
     }
 
     double normalisedStdDev = stdDeviation() / maxAbs;
-    if ( normalisedStdDev < eps )
+    if ( !std::isfinite( normalisedStdDev ) )
+    {
+        // stdDeviation() sums squared values, which can overflow to NaN/Inf for parameter values with
+        // extreme (but finite) magnitudes. Fall back to a range-based variation estimate so the
+        // parameter is not silently treated as having no variation and dropped from correlation
+        // analysis.
+        normalisedStdDev = ( maxValue - minValue ) / maxAbs;
+    }
+
+    if ( !std::isfinite( normalisedStdDev ) || normalisedStdDev < eps )
     {
         return 0.0;
     }