Implementation of multi_image_photometry() code in photometry.py

[JuanCab]

So this is the next big update, I wrote a function multi_image_photometry() that is essentially a wrapper of single_image_photometry(). This will replace photometry_on_directory().

Some changes of note since the last PR:

• Cleaned up single_image_photometry() to
  • output None, None if it fails but it's OK to proceed to the next time. Exceptions are now only used when we really need to stop execution.
  • Fixed logic to find x/y positions from the WCS before attempting to filter out of bounds sources.
  • Check for the presence of require headers in CCDData before attempting aperture photometry.
  • Catch all the repeated warnings thrown by the fitter in compute_fwhm() and just return a summary of the number of failed fits to the screen.
• Fixed compute_fwhm() (really the 2D Gaussian fitter it calls) to handle NaN, which allows us to use NaN for high pixels. (Fixing issue compute_fwhm() throws exception on images with NaN #142)
• I added a test suite that creates multiple CCDData objects with shifted positions to test the new function.

When run on a file in a directory, the output looks something like:

multi_image_photometry: Processing image TIC_467615239.01-S001-R001-C001-ip.fit
  Calling single_image_photometry ...
    > 12 of 926 sources within 2 aperture radii of nearest neighbor ... removed them.
    > 50 sources too close to image edge ... removed them.
    > 864 of 926 original sources to have photometry done.
    > Computing clipped sky stats ... DONE.
    > Fitting FWHM of all sources (may take a few minutes) ... fitting failed on 119 of 864 sources  ... DONE.
    > Bad FWHM values (<1 pixel) for 244 sources.
    > Aperture net counts negative for 1 sources.
    > 245 sources with either bad FWHM fit or bad aperture net counts had aperture_net_cnts set to NaN.
    > Calculating noise for all sources ... DONE.
  Done with single_image_photometry for TIC_467615239.01-S001-R001-C001-ip.fit

I am hoping this is much cleaner and easier to follow without all the warning messages.

[mwcraig]

I think using the ignore_cleanup_errors=False option to TemporaryDirectory will stop the windows failure.

[JuanCab]

Just submitted a commit to add the ignore_cleanup_errors=False to TemporaryDirectory. It seems to have done the trick.

[mwcraig]

There are a couple of minor changes inline.

More broadly, I've opened issues for:

• Figuring out how to better handle messages. We need some way to opt out of (or opt into) t verbose messaging. We might want to implement it using python logging. Logging in the photometry functions #148
• Simplifying the function signatures. Function signatures for photometry functions #149
• Seeing if we can simplify the test image classes a bit. Not a real high priority.

[mwcraig]

Can you add the PR number to this?

Suggested change

	+ multi_image_photometry() now is a wrapper for single_image_photometry() instead of a completely separate function.
	+ ``multi_image_photometry`` now is a wrapper for ``single_image_photometry`` instead of a completely separate function.

[JuanCab]

All issues addressed.