Analyzers reset when changing the ProblemInstance

[pfleck]

When changing the problem-instance of the same problem (e.g. TSP), the Analyzers reset to the default settings.

This happens because different instances sometimes specify different Evaluators, which then fires the Problem_OperatorsChanged, which resets the Analyzers.

This is especially impracticable for creating experiments with different problem instances because the custom analyzers are removed.