WIP: Allow a 2D threshold in StarFinder classes

[krachyon]

As @rainerschoedel noted in https://github.com/astropy/photutils/issues/1284 a non-uniform threshold over an image would be nice to have. As find_peaks already supports this, it appeared to be relatively straightforward to implement in the StarFinder classes.

I won't have much time the next few weeks, so if is this PR is something that you would consider merging let me know and I'll clean it up when I find the time.

What needs a bit of work

• Reading the table in tests with 2D threshold is really hacky.
• A test with an image with variable background to see it in action
• A bit of QOL: Verify the shapes of threshold and data as soon as possible and report understandable error

What I'm not sure about

• Is padding the threshold with zeros the right thing to do? Or should it just be restricted to be used with exclude_border?

[pep8speaks]

Hello @krachyon! Thanks for opening this PR. We checked the lines you've touched for PEP 8 issues, and found:

• In the file photutils/detection/daofinder.py:

Line 415:5: E301 expected 1 blank line, found 0