Adding polar and other plots

[jsub1]

This PR adds the ability to change the projection of the Matplotlib 2D scatter plot. To make sure that selecting ROIs on the various projections, this PR also adds the ability for ROI subset states to apply an arbitrary transform before testing for inclusion in the ROI. There are also lots of tweaks and changes to the scatter code to make sure that things like script export, color-maps, scaling, vectors, error bars, limits, log, tick formatting, etc. behave when using the different projections. A number of tests are added as well. Some future improvements that can be made to this include:

• Curving the theta error bars when using a polar projection.
• Being smarter about units when axes are inherently angular. Currently the code naively assumes that the values of angular axis are in radians. However, if the selected attribute has an attached angular unit, we could convert it before plotting and ROI selection.

[codecov]

Codecov Report

Merging #2170 into master will increase coverage by 0.12%.
The diff coverage is 92.33%.

@@            Coverage Diff             @@
##           master    #2170      +/-   ##
==========================================
+ Coverage   87.86%   87.99%   +0.12%     
==========================================
  Files         246      247       +1     
  Lines       22709    22944     +235     
==========================================
+ Hits        19954    20189     +235     
  Misses       2755     2755              

Impacted Files	Coverage Δ
glue/core/state.py	92.12% <ø> (+0.02%)	⬆️
glue/core/roi.py	91.92% <81.05%> (-1.48%)	⬇️
glue/core/subset.py	89.82% <91.89%> (+0.53%)	⬆️
glue/viewers/scatter/qt/options_widget.py	92.42% <96.55%> (+2.68%)	⬆️
glue/core/roi_pretransforms.py	100.00% <100.00%> (ø)
glue/core/util.py	72.02% <100.00%> (+1.03%)	⬆️
glue/viewers/matplotlib/mpl_axes.py	90.24% <100.00%> (ø)
glue/viewers/matplotlib/qt/data_viewer.py	80.85% <100.00%> (ø)
glue/viewers/matplotlib/toolbar_mode.py	96.11% <100.00%> (+0.11%)	⬆️
glue/viewers/matplotlib/viewer.py	94.30% <100.00%> (+0.37%)	⬆️
... and 17 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update d172634...6988854. Read the comment docs.

[astrofrog]

Fantastic work! This works very nicely for me. There are some small improvements I think we can do (including the ones you mention) but for now I think we can merge and release this.