Code recomendations

[jorgemarpa]

I found some code recommendations/issues while using the first version of tess-asteroids. Some of these are not urgent and can be done later, I add them here to keep track.

• self.shape I was expecting to get the cube shape when calling this. I would also recommend setting shape as an object property

  tess-asteroids/src/tess_asteroids/tpf.py

  Line 122 in 770f46a

  self.shape = shape

• consider target_id or something more explicit than just target

  tess-asteroids/src/tess_asteroids/tpf.py

  Line 36 in 770f46a

  self.target = target

• you could avoid computing the same median twice by assigning to a temporal variable.

  tess-asteroids/src/tess_asteroids/tpf.py

  Line 349 in 770f46a

  bg.append(np.nanmedian(flux_window, axis=0).reshape(self.shape))

• Add/edit relevant keywords in the header, e.g. asteroid target name, CREATOR, PROCVER, TSTART, and TSTOP, etc...

I'll keep reviewing and using the code and add things as they come out in

[jorgemarpa]

Some other ideas to be discussed:

• I notice there are two object attributes, self.ephem and self.ephemeris, the first one has the original DF that the user input to initialize the object, the second one seems to be only the track positions in the frames that fall inside the science portion of the CCD and also match the observing time (by interp). It could be a good idea have a one-to-one mapping between those two, because the new tracks will come with flags per time that tell the quality (interpolation or projection or fitted) of the position.

[altuson]

Something to change in a future PR:

• In the _create_threshold_mask() function, we should add an option to use the predicted asteroid ephemeris as the reference_pixel. This should probably be the default value, instead of center, seeing as the asteroid is not often directly centered.

[altuson]

Something to add in a future PR:

• Create a utils function find_sector(). This would be a wrapper to tess_ephem to find if/when an asteroid was observed by TESS before calling MovingTPF.from_name().

[jorgemarpa]

Other coments for future improvements:

• movingtpf.py is getting too long, 5000+ lines. I think there's a lot of almost repeated code that does the HDU creation. A good idea might be to strip out that code and move to its own fits_utils.py function file. But this could mean some important code modification.
• Move all bspline interpolation to its own function in utils.py (?)