set an infinity translation speed to NaN (will be filled)

[WPringle]

I found a case where a track may have two different locations for the same datetime and produce an infinity speed and hence, generate an error. Corrected to just set the Inf value to NaN where it will be filled by pandas filling operations.

[SorooshMani-NOAA]

@WPringle interesting, can you please add that as a unit test case?

[WPringle]

@SorooshMani-NOAA I was wondering what you think about having an option that tells it to keep the previous 0-hr forecasts (the analysis data) that happened in the past before the forecast_time in the filtered track.data, instead of always filtering those out. i.e., we supply an extra input boolean to VortexTrack like keep_analysis_times.

(What I'm talking about can be seen from the unit test case I just added. )

[SorooshMani-NOAA]

@SorooshMani-NOAA I was wondering what you think about having an option...

@WPringle sorry I missed your comment. I like this idea. The only negative I see right now is that the vortex object is getting a little bit too complicated. In the sense that I need to look up what's happening with different inputs every time I revisit it! I understand the usefulness of having start, end and forecast time as well as the keep 0 hour, but maybe we should just have a plain vortex object and then a couple of filter_*() functions that return filtered objects with certain data removed instead of adding all these different options to the various constructor methods of the class.

With all that being said, for the short term we can add the additional argument, but let's keep the idea of filter functions in mind, so that we can at some point start removing all these stored internal variables and logic and put all into filter methods.

By the way it seems that you missed uploading the track file for the new test. Can you please upload that as well?

[WPringle]

@SorooshMani-NOAA That is an interesting point that since it is all in unfiltered_data then you can just call a filter function to get whatever part you need from it. Yes, let's upgrade that in next version. I'll upload the test data and another test case for this other use case where we append tracks together into one output .dat or .22 file with a custom rename of the advisory column.