A simple way to get closest to both location and time

[jcsmithhere]

It would be good to have a simple way to return bolides close to a point in both time and location. Right now, you have to make one cut first and then the other. For example, say I want the bolides nearest in time to '2022-06-07 22:53:00' but also nearest to the location lon=-31, lat=9. I cannot take each cut separately with a command like this:

bdf_cut = bdf.get_closest_by_time('2022-06-07 22:53:00', n=2)
bdf_cut = bdf_cut.get_closest_by_loc(lon=-31, lat=9, n=2)

That just returns the two bolides nearest in time due to us cutting on time first. I think we need a single method that performs both cuts at the same time.

This would be very helpful if you know the rough time and lat/lon of a specific bolide.

[kaitlynchen1]

I changed the get_closest_by_loc() function to calculate closest based on haversine distance. I also added a new function called get_closest() which uses the following equation:

$$(\frac{\pi \times \textrm{Distance}^2}{\textrm{FOV area}} \times \frac{2 \times \textrm{Time Difference}}{\textrm{Total GLM Viewing Time}}) \times \textrm{(Number of Detections)(Number of Detections-1) / 2}$$

Maybe this calculation will work for the purpose? This method is the same as I use to find bolide cluster pairs, so we are pretending the user input is close to another bolide event in space and time. If you only put in lat/lon then it calls the get_closest_by_loc() function and if the datetime is only entered then it calls the get_closest_by_time() function.

[jcsmithhere]

The above formula is a type of distance metric, but it's really a measurement of how improbable it is two bolides are so close together in time and space. There are generalized distance metrics where you can set the relative weighting between the terms. I showed an example formula near the beginning of this summer.
Although this is a good and useful formula and we can use it here, we should be sure to describe what this formula is actually measuring in the comments and function header. We could then have an option to use a more standardized generalized distance metric. The formula would allow for the user to specify the relative weighting between the terms. Perhaps this would be a second pull request (and it's not necessary for you to write it).