Implement a quick correction for the H4RG persistence

We recently observed bad reduction of H4RG data such as discontinuity between r- and n- and bad sky subtraction in n-arm etc. We suspect but believe that some of them are due to the image persistence of H4RG data. It would be a large amount of work to correct the persistence properly (especially per-pixel correction), but taking into account the demand of science users who already took data in open-use programs and started the analysis, we should provide an easy way to get NIR spectra with an OK level quality.