Changing parameters of some RP leads to wrong reference probability

[mbaudin47]

For example in:
https://github.com/mbaudin47/otbenchmark/blob/52bb104bfa91a7403481e6dc53225be36eecd872/otbenchmark/ReliabilityProblem14.py#L15
the parameters of the distribution or the threshold can be changed. However, this does not update the reference probability. Hence, a user which changes the parameters will have the surprise of getting apparently wrong results.

The situation is correctly implemented in:
https://github.com/mbaudin47/otbenchmark/blob/52bb104bfa91a7403481e6dc53225be36eecd872/otbenchmark/AxialStressedBeamReliability.py#L54

There are several options to implement this.

• Generate an exception if parameters are not left to default. I do not know how to implement this.
• Add a comment in the doc of each problem, to state whether the parameters changes update the reference probability.
• Removing the possibility to change the parameters if the reference probability is not updated according to changes in the parameters. This would restrict the scope of the module, which might seem to restrictive for some users.

[mbaudin47]

Using the maths methods from PR#87 should allow to compute the pf_ref value of some problems depend on the input of the user. This, however, requires to take into account for the threshold, which is not done yet in the scripts. This would involve problems RP14, RP22, RP35, RP75 and FourBranch.