Pytransit

Added

- *TSModel:* An extremely fast transit model for transmission spectroscopy. The model is based on the RoadRunnerModel and can use any
rotationally symmetric function to model stellar limb darkening.
- *LDTkLDM:* A RoadRunnerModel limb darkening model that uses directly stellar intensity profiles created by the LDTk. The model is
parameterised by the effective stellar temperature, surface gravity, and metallicity. This should be quite valuable for transmission
spectroscopy since the number of limb darkening parameters is independent of the number of passbands.

Added

- *EclipseModel:* an easy-to-use secondary eclipse model.

Added

- *OblateStarModel:* transit model to model transits over rapidly rotating gravity-darkened stars by Barnes (2009).
This is an initial release of the model (only a CPU version works at the moment) but I expect to
have it on-par with the rest of the models by v2.4.

Changed

- Renamed the *Swift* transit model introduced in v2.1 to *RoadRunner* transit model.
- Fixed several *RoadRunner* model issues caused by the transition to calculating the
projected distances using Taylor series expansion.
- Changed several of the OpenCL models use the Taylor series expansion approach to calculate
the projected distances.
- Lots of minor bug fixes.

PyTransit version 2.2 now calculates the normalized planet-star distances using a Taylor series expansion
of the x, and y positions in the sky plane (Parviainen and Korth, 2020, submitted to MNRAS). This gives a
significant speed boost in transit model evaluation that is especially noticeable for eccentric orbits.

PyTransit version 2.1 adds a new transit model named *swift* that can use any Python callable to model the stellar
limb darkening while still giving equal or better performance than the analytical quadratic transit
model.

Added

- *Swift* transit model (Parviainen, submitted) to allow fast and flexible transit modelling with
any radially symmetric limb darkening model.

PyTransit Version 2 removes the Fortran dependencies in v1 by implementing all the transit models
in *numba*-accelerated Python. Version 2 also adds a number of new transit models, and implements
most models both in CPU and GPU versions.

Added
- New API that is consistent across all transit models.

Changed
- Nearly everything.