0.3.0b8 json gen adjs

[carueda]

No description provided.

[carueda]

change resulted in a reduction of the time to write json files from about 100 minutes to... about 30 seconds

The files look good by spot check, and they have all the same file sizes as the earlier run that had redundant creation of the dataframe. I will proceed with checking full temporal coverage, however I do see one issue just from examining the json files -- There are extra input flac files in some of the daily json files. We need at most one file from the previous day to cover the beginning of a day, however series of dates include two files from the previous day, one that definitely does not have data for the day being processed. I would guess it is caused by a fixed buffer definition for tapping into files that began the previous day, rather than finding the first file that has data for a day. It probably does not affect processing, so we can choose to ignore it, but I do think we ought to be able to code for correct constraint of only files that contain data for a day. For the record, the dates in 2020 with the extra files are 2/10 through 3/13, and 12/10 through 12/31.