Handle missing values correctly

[felixcremer]

I fear we are not handling the missing values correctly in the construction of the data cubes and therefore it might be, that we are using a wrongfully filled time series for the RQA analysis.

[felixcremer]

This was not fixed in #51 because the problem is at the construction of the data cube not in the handling of the missing values in the rqatrend function. We might for some data cubes not have missing values in the inner function of rqatrend, because we are not correctly converting missing_value data to missing.

[felixcremer]

I hope that this is fixed in DiskArrayTools 0.1.11 but we would have to test that fix here.