Reproducibility of ensemble replay settings (possible bug)

[rtodling]

Wei reports to me that he has a case where the ensemble replay (of x0048) does not reproduce the original (x0048). I have not tested to confirm. I am putting this here so it doesn't get forgotten.

[gmao-wgu]

@rtodling If my exp starts from 20220101_15z, the analysis cannot be reproduced at 20220103_12z. If my exp starts from 20220103_03z immediately prior to 20220103_12z, the analysis at 20220103_12z can then be reproduced, but it will fail to reproduce at a different later synoptics time. Because I already deleted that particular run, I could not recall at what time the analysis can not be reproduced.