Asilib

Added
- RGB auroral intensities in `asilib.Conjunction.intensity()`.
- An auroral intensity test (from the nearest pixel and equal area) using the TREx-RGB data.
- `asilib.skymap.geodetic.skymap()` function that maps the (az, el) skymaps to (lat, lon) skymaps assuming Earth is a sphere (i.e., not an ellipsoid).
- Added a plot test comparing the official and the asilib THEMIS GILL (lat, lon) skymaps.

Changed
- The `asilib.Conjunction()` class had am ambiguity regarding whether if the satellite ephemeris was interpolated (or downsampled to) the ASI time stamps, or kept at the original cadence. This ambiguity made calculating auroral intensity error-prone, so now `asilib.Conjunction.intensity()` automatically interpolates the satellite ephemeris.

Fixed
- A bug in the asilib.Imager.data property that relied on hard-coded filtering of unfilled images using `np.where()`. This led to duplicate time stamps for RGB images, and a crash with `asilib.Conjunction.intensity()`.
- Removed the THEMIS and REGO imports at the top-level of asilib, i.e., in `asilib/__init__.py`. Now they are imported in `asilib/asi/__init__.py`, so users must always import asilib.asi when using ASI modules contained within the `asi/` folder.

Added
- The TREx-RGB loader, `asilib.asi.trex_rgb()`, courtesy of C. McKenna.
- `asilib.asi.trex_rgb()` tests
- `Conjunction.lla_footprint` tests
- `asilib.asi.trex_rgb()` to the online documentation.

Fixed
- A bug in `Imager.keogram()` that calculated the incorrect pixels that is used for slicing images when assembling a keogram. While this bug does not affect ASIs with square pixel resolution, it did for TREx-RGB which is rectangular. The fix slightly modified the keograms, by about a pixel, so I regenerated the baseline keogram plots.
- A bug in `Conjunction.lla_footprint` where the `alt` variable was rewritten. As a result, IRBEM did not map to the requested altitude.

Added
- `asilib.Imagers.get_points()` method to get the (lon, lat) and pixel intensity points.
- Tests for `asilib.Imagers.get_points()`

Added
- `asilib.Imagers()` class to make mosaics and synchronize plotting multiple imagers. The first two methods implemented are:
- `asilib.Imagers.plot_fisheye()` to plot fisheye lens images from multiple imagers, and
- `asilib.Imagers.plot_map()` to project images from multiple imagers onto a map. This method has an `overlap` kwarg that by default masks out the (lat, lon) skymaps that map to pixels that overlap with a neighboring imager. This is a surprisingly efficient method that needs to be run just once (tested for up to five imagers; I'm unsure how a 10 or a 20 imager array will fare).
- Added plot tests for the two `Imagers` methods.
- Added an `Imagers()` description in the Get Started and API pages.
Changed
- Refactored the Donovan+2008 auroral arc example to use `asilib.Imagers.plot_map()`.
- Added a check to allow custom `matplotlib.colors.Normalization` object to be passed for `color_norm`.

Added
- Support for RGB images in `asilib.Imager.keogram()`.

Changed
- Simplified `asilib.Imager.__getitem__()` to repay the technical debt. Before the method was difficult to reason about and it processed the [time] and [start_time:end_time] slice cases separately. Now `__getitem__` handles both of those cases in the same way.
- Simplified how the `Imager._keogram` array is allocated when `asilib.Imager.keogram()` is called.
- Edited the Imager flowchart.

Added
- Deprecation warnings to the legacy asilib plotting functions and in the API reference. They will be removed in or after December 2023.