Rm-tools

-updates to do_fitIcube tool: removed deprecated cutoff keywords, fixed tqdm
progress bars, some minor speed and efficiency improvements.
-relaxed numpy version restriction, since the numpy-astropy version conflict
has been resolved.
-added tqdm as install dependency.
-removed internal copies of external packages (corner, emcee, nestle) used by QU-fitting
-added pre-commit auto-formatting system
-escaped all LaTeX characters
-implemented auto-formatting, using black, and updated all existing code to that standard.
-removed RMS noise estimate for FDF -- it was inacurrate and misleading
-fixed corMAD noise estimate for FDF -- it was estimating MAD of PI rather than Q and U,
and thus was biased low. Now uses Q and U of FDF, and reproduces correct value in
signal-free and RM-cleaned simulations. Still biased high in dirty spectra; no fixing that.
-old versions of core rmsynth and RMSF calculation functions are now officially deprecated
-fixed crashing when trying to generate RMsynth1d plots for source with negative Stokes I model
-improved FITS headers of 3D tools output
-1D: removed RMS noise estimate (for being misleading); removed _chan outputs (for being low value and confusing)
-fixed estimation of empirical FDF noise. (1D RMclean)
-QU-fitting: fixed problems with de-rotated angle wrapping
-Integrated nufft tool for the Fourier transform math (makes rmsynth faster)
-added options to rmsynth3d to decrease RAM use
-new QU-fitting models: m5 (single Burn slab), m6 (double Burn slab), m7 (single mixed emission slab)

-progress bars updated to use tdqm, which should make them faster.
-Integrated super-resolution method of Cotton & Rudnick 2023, which creates
narrower RMSF shapes by setting lambda^2_0 = 0.
-Added options to dofitIcube tool to save more outputs (coefficient maps,
errors, etc).
-Made RMSF theoretical width calculations consistent: Dickey et al. 2019
now used throughout.

-merged in bwdepol branch. This does not change any core functionality in the
existing RM-Tools programs, but adds two new command-line tools: rmtools_bwpredict
and rmtools_bwdepol, which predict and correct for bandwidth depolarization
respectively.
-fixed a bug with JSON outputs not handling numpy variables. This was partially
due to a problem with asscalar() in the utils using a depcreciated numpy function,
but partially from other causes.
-an improved version of the Stokes I cube fitting routine has been developed
by Lerato Sebokolody, which includes parallelization.

-changed Stokes I fitting to always use 64 bit values (even if data is 32 bit).
This is because the fitter was found to be vulnerable to numerical issues
when fitting log-polynomials in 32-bits (thanks to Shannon Vanderwoude for
discovering the problem).
-added parameter errors of Stokes I fit as output to Stokes I fitDict, to
RMsynth1D outputs, and QU-fitting outputs. These are the 1-sigma errors
computed by the mpfit routine.
-the renormalize_StokesI_model function is now depreciated, since it cannot
transform the errors. It is left in for the time being in case it is needed
for any legacy reasons.
-the make_freq_file.py/rmtools_freqfile tool was found to sometime produce
truncated output (not all channels included). This was fixed and changed
to a more rigourous calculation (using WCS, as suggested by Alec Thomson).
-modified do_fitIcube.py/rmtools_fitIcube to output maps of the fit coefficients
and their errors. Thanks to Biny Sebastian for providing this code.
-merged pull request from Alec Thomson that changes the fractional polarization
reported by rmsynth1d to use the bias-corrected polarized intensity.
!!!Previous versions used the non-corrected polarized intensity, from this
version onwards the corrected polarized intensity will be used!!!
-merged pull request from Alec Thomson that implements windowed cleaning
and adds a command line flag to set the windowed clean threshold.
-merged pull request from Alec Thomson that changed the QU-fitting routine
to use the bilby package for the nested sampling. This package allows the
user to specify which sampler they wish to use -- the default is dynesty,
but users can choose to use pymultinest or other supported samplers.
-removed support for Python 3.6, as a consequence of the change to bilby.
From this point forward, Python 3.7 is the minimum version supported
(at least for the QU-fitting portion of the package).

-modified the generation of maxPI and peakRM maps in RMsynth3D so that
it occurs in its own function rather than as part of the file writing
function; this allows it to be run inside a notebook.
-modified RMpeakfit_3D to be more stable when FDF cubes have NaNs present.
-Stokes I fit coefficients no longer become NaNs when negative Stokes I
models are renormalized to a new reference frequency.
-added test for unreasonably large peak RM (compared to computed range)
as this can happen when peak is on edge of spectrum. Changes RM to NaN in
these cases.
-Fixed crash when sigmaAdd calculations received NaNs in inputs.
-added experimental tool to compute QU models from clean components,
requested by Luke Pratley. This currently has a normalization error of some
kind, which I've asked Luke to chase down.
-dropping support for Python3.5, since the QUfitting code has problems with it
and it's already end-of-life.
-improved the calculate_RMSF script to not crash when given width smaller than
first sidelobes. Also corrected it to produce the correct width: previously
when width was specified by user it would be 2x wider than specified.
-fixed problem with do_fitIcube.py which would crash when fed 4D Stokes I cube.
-fixed undesired behaviour in rmsynth1d and rmclean1d where spectra plots
wouldn't be saved when plotting to screen was turned on.

-updated calculate_RMSF to use same theoretical width as rmsynth1D
(the formula from Dickey et al. 2019)
-fixed problem where very small Stokes I model values (~1E-60) would
cause numeric overflow in Q/I, U/I, creating infinities that were
not handled gracefully. Now converting infinities to NaNs.
-fixed peakfitcube not being callable from the command line.
-fixed typos in rmtools_calcRMSF.
-fixed pip installation so that QUfitting models are correctly installed.