A couple of things to fix on the remote data fetch:
Better handling of the half-written file at the top of the catalog. We should check the size of the file on the disk vs the listing in the catalog to see if it was truncated.
Multiple downloads -- if the analysis is aborted, the data flow stops. We need to play catch up during a restart.
Out of order fetches -- the program appears to go backwards in time if new data is grabbed from before. We need to deal better with negative extrapolation times.
A couple of things to fix on the remote data fetch: