Added fixes for some bugs in the to_fits methods of LightCurve classes

[christinahedges]

This PR fixes bugs in the LightCurve, KeplerLightCurve, and TESSLightCurve to_fits methods and the tests. This ensures users can pass extra keyword arguments.

@Nschanche points out that it's odd that, by default, to_fits doesn't save every column. Some of the columns can be a bit superfluous, so I'm not sure of a neat way forward here. What do people think, should to_fits save all columns by default?

[christinahedges]

@Nschanche when you review this you'll see I've accidentally linted this file....but I don't know that I hate that...

[Nschanche]

@Nschanche points out that it's odd that, by default, to_fits doesn't save every column. Some of the columns can be a bit superfluous, so I'm not sure of a neat way forward here. What do people think, should to_fits save all columns by default?

There are a number of keywords that are dropped by to_fits as it is now, including timecorr, sap_flux, sap_bkg, sap_bkg_err, psf_centr1+2 and errors, and pos_cor1+2 (the last two are addressed in this pull request). I tend to think everything should be saved by default, but we could make it an explicit choice by, e.g., adding a flag such as all=True that would save every column, but stick with the current subset by default.