v1.4.2.0 fix gravgroup, extend HasEclipticLongitude

• Fix edge case in grav group: incorrect casting in the C code was causing planets that were too close to a sector boundary or another planet to be thrown back into the first sector.
• Remove cuspsToSectors.
• Add ability to set longitude in HasEclipticLongitude typeclass

What's Changed

• Fix gravgroup by @lfborjas in #41
• 1.4.2.0 by @lfborjas in #42

Full Changelog: v1.4.1.0...v1.4.2.0

V1.4.1.0 Minor fixes, confirm support for GHC 9.2.1

• Export utcToJulianDays, to obtain a product of (TT, UT1) Julian Days from a UTCTime value --
  saves you one IO trip vs. getting them separately.
• Support for GHC 9.2.1
• Fix minor bug in directionChange that made lookups starting <=30 minutes before the event fail
  (due to an artifact of the original C function that explicitly excludes such lookup bounds.)

v.1.4.0.0 -- Time refactoring, crossings, phenomena

BREAKING CHANGE: Major refactoring of time values: JulianDay is no longer an alias for Double,
and is now a type that carries a witness of its provenance -- which enables us to work with functions that
transact in both Terrestrial (Ephemeris) Time, and Universal Time, without mixing them up.

• Upgrades to version 2.10.02 of the C library.
• Adds the Precalculated namespace with functions to read and write pre-calculated
  ephemeris from a file on disk. Useful when examining an interval of time for ecliptic phenomena.
• Introduces the SwissEphemeris.Time module, with various conversions between time standards
  and some Haskell time values.
• Introduces functions that find moments of exactitude for longitude crossings (heliocentric and geocentric,)
  and moments of exactitude for lunar phases. The geocentric interpolation, as well as lunar phases,
  use the Brent-Dekker algorithm for root finding, the heliocentric functions are able to do faster
  parabolic approximation. I need to study the C sources a bit more to see if something similar is possible
  for interpolation in general (see the events files in the C sources for ideas.)
• Adds functions for calculating planetary phenomena as visible from earth, as well as solar and lunar eclipses.
• [dev] adds a Dockerfile and NOTES.md to aid in diagnosing memory leaks.
• [dev] adds some basic Nix derivations for producing documentations and a release tarball.

Minor pointer hygiene and error reporting update

• More memory safety paranoia: ensure that gravGroup functions have appropriately
  scoped "extraData" pointers; was using an unsafe function before that somewhat inexplicably never failed, except for one time maybe while doing unrelated testing.
• Fix empty sector handling for gravGroup: it now correctly reports that zero-length
  sectors are not supported

Fix error string allocation, bundle fixtures for tests

A couple of memory safety patches:

• Attempt to rein in memory unsafety by keeping all pointer peeking in IO for gravGroup fns.
• Always allocate 256 chars for error messages.
• [dev] Bundle test ephemeris into the hackage tarball, to allow hackage CI and nixOS to
  successfully run tests.

Upgrade swiss ephemeris to 2.10.01, include gravgroup

• Drops support for base < 4.10, which effectively excludes GHC versions less
  than or equal to 8.0.2. Supporting older haskells gets more cumbersome with each
  addition. Please submit a patch with all the preprocessor magic if you really
  want old haskell support.
• Add ChartUtils namespace, with a couple of convenience functions for chart drawing.
• Upgrade to swiss ephemeris 2.10.01.
• Update LICENSE to AGPL -- it changed in the C library, too.
• [dev] Fix QuickCheck test dependency, to hopefully be copacetic in NixOS builds.
• [dev] add optional nix scripts.
• [dev] a whole buncha hlint + autoformat fixes.

Update base, now compatible with ghc 8.10

Merge pull request #21 from lfborjas/update-base

Allow newer haskells, add to testing matrix.

Adds `gregorianDateTime`, reverse of `julianDay`

See CHANGELOG.md

Better `splitDegrees`, exposed time constructors, Nakshatras

BREAKING CHANGE: splitDegrees now takes options that reflect the options in the underlying library.

• Constructors for JulianTime and SiderealTime are now exposed.
• Introduce SplitDegreesOption enum for all options one can split degrees with; amends splitDegrees to take
  said options as the first argument.
• splitDegreesZodiac is unchanged, though a mere veneer for the now more powerful splitDegrees.
• Since splitDegreesZodiac goes the extra enum-mile to provide human-readable zodiac names, and the underlying
  library can also split on Nakshatras, we now include the NakshatraName enum. Names are from wikipedia
  and I saw some variants, so please forgive any mispellings!

Better position types, newtypes for time, more position calculations, some utils

BREAKING CHANGE: the Coordinates type has been retired, in favor of the more specific
GeographicPosition and EclipticPosition. calculateCoordinates is now calculateEclipticPosition,
and the calculateCusps* family now takes a GeographicPosition as part of its inputs.

• Introduces an Internal module with types and helpers that this library introduces,
  which are not native to the underlying C library. Import at your own risk! (the "curated"/
  "stable" ones are re-exported by the main module.)
  • Deprecates the Coordinates type, in favor of EclipticPosition.
• Introduces functions to calculateEquatorialPosition and calculteObliquity at a given time,
  as well as types that better convey the different types of positions (EquatorialPosition, ObliquityInformation).
• Some astrology helpers: convert between equatorial and ecliptic (and vice-versa,)
  obtain the Delta Time effective at a given moment, obtain the house position of a given body.
  (Note: the calculateHousePosition function is more useful for working near the polar circles or for bodies
  off of the ecliptic -- the ARMC and obliquity need to be calculated or provided, it's simpler
  if you already have the cusps: just check which cusps a given longitude falls between -- no need for
  this function!)