This PR does a few improvements to make the package more generalizable.
Move all stan related packages to "Enhances". Currently rstan is in "Suggests", but I don't see why rstan should be treated differently
Move the basic functions related to brmsfit objects to function-specific folders just like other stan interfaces. Only functions solely used for brmsfit objects stay in a separate file
Adds better support for target_observation_weights to make it easier to do loo calculations.
Adds tests for using target_observation_weights with different stan objects.
TODO:
Add tests also for cmdstan objects
Implement cmdstanfit and stanfit methods so that they also update & return the model object.
Make all tests work again
Come up with a better way to easily handle expectations computed using constrained parameters with brms, rstan & cmdstanfit without user having to do anything special.
This PR does a few improvements to make the package more generalizable.
target_observation_weights
to make it easier to do loo calculations.target_observation_weights
with different stan objects.TODO: