Consider dropping M0 (mean anomaly at the epoch) in favour of λ0 (mean longitude at the epoch)

[j-faria]

[ThomasBaycroft]

I have implemented in the BINARIESmodel a flag "use_binary_longitude" which allows to switch from sampling in M0 to sampling in lambda0.

This doesn't change the output which is still phi0(/M0) so that plotting is not broken, but the at the stage of the Keplerian w is subtracted from phi, such that what is being sampled in (and therefore what the prior is defined on) is actually lambda0.

[j-faria]

It might still be a good idea to use this parameterization for other models