Duplicate instances of `self._setup_complex_phase()` in relative_setup.py

[zhang-ivy]

I noticed that self._setup_complex_phase() is called twice.

• first time: https://github.com/choderalab/perses/blob/main/perses/app/relative_setup.py#L333 -- as noted in the comments, this call needs to happen before the system generator is created because for host guest systems, the host molecule needs to be supplied to when creating the system generator
• second time: https://github.com/choderalab/perses/blob/main/perses/app/relative_setup.py#L409 -- not sure if this is necessary, try removing it and seeing what happens?

[zhang-ivy]

Iván notes that this is not super high priority since we are in the midst of overhauling the API anyway

[ijpulidos]

Yes,the comment on

perses/perses/app/relative_setup.py

Line 329 in ae55801

# We must generate the complex topology and positions before the SystemGenerator is created because if the

explains why we needed that at that instance but, as discussed, we can quickly investigate what happens if we remove the second call, since that seems redundant and it shouldn't be doing anything new.

As a reference, when I worked in the refactor for the NEQ cycling protocol, the RelativeFEPSetup.__init__ method of is more organized and it's only setting up the phases at the end of it, as per

perses/perses/app/relative_setup.py

Line 349 in faac030

self._setup_phases(

(just to keep in mind).