Astrobase

Fixes

- lcproc: fixed various errors and typos
- checkplotserver: fixed not being able to specify < 2 periodogram peaks in
period search tab
- checkplot: fixed handling of SNR keys from newgen checkplots

New stuff

- checkplot: can now provide epochs for folding phased light curves as input,
overriding the default light curve time-of-minimum finding strategy
- starfeatures and checkplot: can now use arbitrary mags and colors in input
objectinfo dicts, can add custom bandpasses as well
- checkplotserver UI: added highlight for each GAIA neighbor in the DSS finder
chart as cursor mouses over its row in the GAIA neighbor list table to help
with visualizing close neighbors
- tests: added tests for new epoch and mag/color handling
- added a CITE.md


Changes

- checkplot: no longer warns if a pickle was written in Python 2 and latin1
encoding was needed to read it in Python 3
- checkplot: a Savitsky-Golay fit is now tried before giving up on finding
time-of-minimum of a light curve if the initial spline fit failed
- checkplotserver: UI fixes to not squish phased LC tiles into too small an
area; now they overflow off the page with scroll-bars, but should remain
legible
- `lcmath.sigclip_magseries` can now use either median/MAD or mean/stdev as the
central and spread values in determining where to clip (added by
joshuawallace in 40)


Fixes

- checkplotserver: handle more nan-JSON breakage
- starfeatures: handle missing information correctly
- fixed link at the top of hatlc.py to the astrobase-notebooks repository (added
by adrn in 37)
- checkplotlist: fixed to not use non-existent `str.isdecimal()` in Python 2
- various other Python 3 specific things were fixed to make them work on python
2.7 as well

Fixes

- astrokep: fix EPD application
- checkplot.cp2png: guard against missing phased LCs for nbestperiods
- checkplot: fix docstring
- cp: multiple invocations of the same period-finder method allowed now
- cpserver: handle previous-gen checkplot pickles correctly
- cpserver.js: better handle missing checkplot info from GAIA, etc.
- imageutils: some more tools added
- lcmath: add `sigclip_magseries_with_extparams`
- periodbase: periodepsilon now a fraction diff instead of actual diff
- `starfeatures.neighbor_gaia_features`: handle missing K mag
- test_lcfit: fix expected val for weird tiny diff in fit coeffs on test runner
- tests: added test for 34
- tests: various fixes
- varbase/trends: added implementation of classic EPD and random-forest EPD

Fixes

- kbls: fixing nphasebins and messaging about autofreq
- lcfit: add legendre fit coeffs to output dict
- test_lcfit: rtol fix
- tests: added lcfit, periodbase tests, `test_endtoend` -> `test_checkplot`

Fixes

- imageutils: some reorganization
- kbls.bls_snr: fix for not using the same kwargs from the initial BLS run
- lcproc.filter_lclist: get the whole matching row for xmatches
- MANIFEST.in: ignore JPL ephemerides too
- setup.py: use pip>=6 req format, fix conditional py2 dep on future pkg
- timeutils: use HTTPS url for JPL ephem download

Fixes

- MANIFEST.in: removed pngs from examples to cut down on dist size