ClassNames caching leads to problems when dataset is not cloned

[abeham]

When a DataAnalysisSolution is created with the same ProblemData instance that the user edited, then the wrong class names will be used.

[mkommend]

r8554:

• Corrected class names and class values caching in ClassificationProblemData
• Removed caching of classification penalties
• Corrected AccuracyMaximizationThresholdCalculator (retrieving of penalties was switched)
• Added asserts for the achieved accuracy to the classification sample unit test
• Adapted ClassificationEnsembleSolutionEstimatedClassValuesView to the changes in the ClassificationProblemData

[abeham]

It does not work.

I opened the classification sample, let the algorithm run a couple of generations, reset it and renamed the class names. Then let the algorithm run again. It still says "Class 0" and "Class 1" in the charts. I also tried creating a new algorithm and loading a new problem, to no avail. Setting the class names doesn't have any effect.

[mkommend]

I could reproduce the issue and corrected it with r8716 (Corrected constructor of classification problem data to take class names into account).

[mkommend]

Forgot to adapt display in classification penalties views.

[mkommend]

r8717: Corrected class names in classification penalties views. Additionally I corrected a bug regarding the copy constructor, as the penalties were reset before.

[mkommend]

r8745: Overlooked a minor bug in the classification problem data regarding the copy constructor.

[mkommend]

r8802: Corrected multiple event registration in ClassificationProblemData.

[abeham]

I tested it and it works