Hi, I have a general and easy question. The documentation for data input states a long format; due to the data structure in the preprocessed data, I have used what I would call a wide format. The original data has columns: unique_id, ds, observation_type, value, test_group. There are 50 different observation_types. Pivot to wide format leads to the 50 individual observation_type columns where the sparse data becomes a problem (the reason I like the idea of using long format), I also reduced the DateTime to days, which causes loss in temporal resolution, as the sampling frequency of some observation_type's is high others low. Some observation_types, as temperature, were measured several times a day in some cases. So I am unsure how to correctly preprocess the multivariate input data if there is a Nixtla method that could handle it as a long format. The long format is usually only applicable to univariate time series datasets. If someone could clarify it I would appreciate it