Avoid setting all end-state properties when decoupling

[fjclark]

This closes #252. BSS.Align.decouple() no longer sets LJ1, element1, or ambertype1, rather this is done by the _to_pert_file method of Process.Somd. The related tests of "decouple" have been removed, and direct tests on the SOMD pert files have been added.

• I confirm that I have merged the latest version of devel into this branch before issuing this pull request (e.g. by running git pull origin devel): y
• I confirm that I have permission to release this code under the GPL3 license: y

Suggested reviewers:

@lohedges, @xiki-tempula

[xiki-tempula]

LGTM

[lohedges]

Looks good. The test failure on Python 3.11 was because of pytest-black. It appears to be unhappy with the LaTeX formatting in one of your matplotlib commands. Could you test this locally to see if there's a workaround?

[fjclark]

Thanks for the comment @lohedges. My understanding was that the error was due to formatting in _somd.py, and that the matplotlib issues were just warnings (from files that I haven't changed). Is there an error message I've missed for the matplotlib commands?

I'm confused, because the error message I could see seems to show that black is unhappy with an extra line:

 --- $SRC_DIR/python/BioSimSpace/Sandpit/Exscientia/Process/_somd.py	2024-02-26 19:31:10.766528+00:00
+++ $SRC_DIR/python/BioSimSpace/Sandpit/Exscientia/Process/_somd.py	2024-02-26 19:36:43.213044+00:00
@@ -1032,11 +1032,10 @@
     mol = molecule._sire_object.__deepcopy__()
 
     # If the molecule is decoupled (for an ABFE calculation), then we need to
     # set the end-state properties of the molecule.
     if molecule.isDecoupled():
-
         # Invert the user property mappings.
         inv_property_map = {v: k for k, v in property_map.items()}
 
         # Get required properties.
         lj = inv_property_map.get("LJ", "LJ")

but I auto-format with black on saving, and running black directly on the file left it unchanged. Could a different version of black be being pulled in here?

To get round this, I've just removed the line.

[lohedges]

Apologies, I misread the error message. You are correct, the matplotlib stuff is just a deprecation warning. The tests are currently run using black 23. Perhaps you have a different version locally. Deleting the is the correct fix.

[xiki-tempula]

@lohedges Do you mind port this to the main as well, please? Thank you.

[lohedges]

Yes, of course. I was planning on doing that today but didn't get time. Will do it first thing tomorrow.

[xiki-tempula]

Thanks