Only print constraint warning once

[lohedges]

This PR adds a small fix to ensure that the asymmetric constraint warning is only printed once, i.e. it isn't duplicated if multiple constraints don't match between end states. I've also added an initial check to make sure that the number of constraints is the same, then check each constraint one-by-one.

[lohedges]

I'm confused by the CI errors for macOS. I had assumed that this would be fixed when the latest Sire and BioSimSpace packages were built.

  Could not solve for environment specs
  The following packages are incompatible
  └─ biosimspace is not installable because there are no viable options
     ├─ biosimspace [2023.1.0|2023.1.1] would require
     │  └─ sire ~=2023.1.0 , which does not exist (perhaps a missing channel);
     ├─ biosimspace 2023.1.2 would require
     │  └─ sire ~=2023.1.3 , which does not exist (perhaps a missing channel);
     ├─ biosimspace 2023.2.0 would require
     │  └─ sire ~=2023.2.0 , which does not exist (perhaps a missing channel);
     ├─ biosimspace 2023.4.0.dev would require
     │  └─ sire 2023.4.0.dev , which does not exist (perhaps a missing channel);
     ├─ biosimspace 2023.5.0.dev would require
     │  └─ sire 2023.5.0.dev , which does not exist (perhaps a missing channel);
     └─ biosimspace 2024.1.0.dev would require
        └─ sire 2024.1.0.dev , which does not exist (perhaps a missing channel).

Perhaps this runner is still running macOS x64, not arm64?

[lohedges]

I also don't understand this bit:

  Warning: 
  LibMambaUnsatisfiableError: Encountered problems while solving:
    - nothing provides sire ~=2023.1.0 needed by biosimspace-2023.1.0-py38_0

We are explicitly using py311 in our CI build matrix, so why is it complaining about a dependency for py38? There are similar errors for Windows:

 Warning: 
 LibMambaUnsatisfiableError: Encountered problems while solving:
   - nothing provides sire ~=2023.1.0 needed by biosimspace-2023.1.0-py38_0
 
 Could not solve for environment specs
 The following packages are incompatible
 \u251c\u2500 biosimspace is installable with the potential options
 \u2502  \u251c\u2500 biosimspace [2023.1.0|2023.1.1] would require
 \u2502  \u2502  \u2514\u2500 sire ~=2023.1.0 , which does not exist (perhaps a missing channel);
 \u2502  \u251c\u2500 biosimspace 2023.1.2 would require
 \u2502  \u2502  \u2514\u2500 sire ~=2023.1.3 , which does not exist (perhaps a missing channel);
 \u2502  \u251c\u2500 biosimspace 2023.2.0 would require
 \u2502  \u2502  \u2514\u2500 sire ~=2023.2.0 , which does not exist (perhaps a missing channel);
 \u2502  \u251c\u2500 biosimspace 2023.4.0.dev would require
 \u2502  \u2502  \u2514\u2500 sire 2023.4.0.dev , which does not exist (perhaps a missing channel);
 \u2502  \u251c\u2500 biosimspace 2023.5.0.dev would require
 \u2502  \u2502  \u2514\u2500 sire 2023.5.0.dev , which does not exist (perhaps a missing channel);
 \u2502  \u251c\u2500 biosimspace 2024.1.0.dev would require
 \u2502  \u2502  \u2514\u2500 sire 2024.1.0.dev , which does not exist (perhaps a missing channel);
 \u2502  \u2514\u2500 biosimspace 2024.2.0.dev would require
 \u2502     \u2514\u2500 python_abi 3.12.* *_cp312, which requires
 \u2502        \u2514\u2500 python 3.12.* *_cpython, which can be installed;
 \u2514\u2500 python 3.11**  is not installable because it conflicts with any installable versions previously reported.

The conda environment should be created using ${{ matrix.python-version }}. This is what we do for Sire and BioSimSpace and don't have any similar issues there.

[mb2055]

All good (apart from the CI issues) 👍

[lohedges]

For macOS I see the following:

Platform: osx-64

Whereas for Sire and BioSimSpace it's:

Platform: osx-arm64

I'm not sure why we repeatedly get osx-64 runners for this repo, whereas the others are now osx-arm64. Both specify macos-latest in the CI config.