Updating the AIMNet package

[cplett]

This PR updates the AIMNet2 integration to use the currently maintained package (aimnetcentral), enabling access to all supported models.

Closes #32.

[stgeo]

Thanks - looks much better now!

In addition to the inline comments, I noticed two things:

• I got a PyTorch/CUDA warning:

venv-aimnet2/lib/python3.12/site-packages/torch/cuda/__init__.py:371: UserWarning: Found GPU0 NVIDIA RTX A500 Laptop GPU which is of compute capability (CC) 8.6.
The following list shows the CCs this version of PyTorch was built for and the hardware CCs it supports:
Please follow the instructions at https://pytorch.org/get-started/locally/ to install a PyTorch release that supports one of these CUDA versions: 12.6, 12.8, 13.0
  _warn_unsupported_code(d, device_cc, code_ccs)

It goes away with -d cuda. Not sure how to make it not complain. In any case, it's just a warning and it works correctly on both GPU and CPU.

• The test references need to be updated for the new default model:

======================================================================
FAIL: test_H2O_engrad (__main__.Aimnet2Tests.test_H2O_engrad)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/tests/aimnet2/test_aiment2_standalone.py", line 58, in test_H2O_engrad
    self.assertAlmostEqual(energy, expected_energy, places=9)
AssertionError: -76.4768264497 != -76.47682538331 within 9 places (1.0663899985274838e-06 difference)

======================================================================
FAIL: test_OH_anion_eng_grad (__main__.Aimnet2Tests.test_OH_anion_eng_grad)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "tests/aimnet2/test_aiment2_standalone.py", line 93, in test_OH_anion_eng_grad
    self.assertAlmostEqual(energy, expected_energy, places=9)
AssertionError: -75.82629740302 != -75.82629634884 within 9 places (1.0541800037344728e-06 difference)

======================================================================
FAIL: test_OH_rad_eng_grad (__main__.Aimnet2Tests.test_OH_rad_eng_grad)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "tests/aimnet2/test_aiment2_standalone.py", line 128, in test_OH_rad_eng_grad
    self.assertAlmostEqual(energy, expected_energy, places=7)
AssertionError: -75.68258800204 != -75.68258695326 within 7 places (1.0487800068403885e-06 difference)

[cplett]

@stgeo regarding the warning: It seems to me that the cuda version you have installed does not match your GPU. So this is probably a warning the user might want to have. Otherwise, we could also filter it with warnings.filterwarnings.

The tests are updated now.

[stgeo]

Thanks, good to go!