Describe the bug
(Now edited for clarity)
I've found an interesting issue with some chiral-at-P compounds. It comes in two, presumably related, parts, which I'll illustrate with the model compound with SMILES C[P@]1[C@H](C)CCC1:

1. rdkit represents this SMILES with the methyl groups syn to each other, whereas they should in fact be anti. I read the SMILES documentation and just wound up confused, in the context of this molecule...
2. I then looked at the CIP assignments. rdkit describes the syn isomer as (R,R)-, but the P stereocentre should be (S). This remains true when using the more accurate FindPotentialStereo() approach, as kindly suggested by Greg Landrum.

This image, in which ChemDraw is correct both times, might make both of the above clearer:

To Reproduce

from rdkit import Chem

smiles = r"C[P@]1[C@H](C)CCC1"
mol = Chem.MolFromSmiles(smiles)

# Assign stereochemistry to stereocentres
potential_stereo = Chem.FindPotentialStereo(mol)
Chem.AssignStereochemistry(mol, cleanIt=True, force=True)

for centre in potential_stereo:
    atom_idx = centre.centeredOn
    atom = mol.GetAtomWithIdx(atom_idx)
    if atom.HasProp('_CIPCode'):
        cip = atom.GetProp('_CIPCode')
        atom.SetProp('atomNote', cip)

Expected behavior

1. rdkit should have created the molecule with opposite configuration at the phosphorus stereocentre.
2. rdkit should have assigned the phosphorus stereocentre of the molecule it drew oppositely in the CIP system. The CIP assignment did match the SMILES, but was incorrect for the molecule that was incorrectly drawn.

Configuration (please complete the following information):

• RDKit version: 2024.3.5
• OS: Windows 10 Enterprise 10.0.19045
• Python version (if relevant): 3.12.4
• Are you using conda? No
• If you are using conda, which channel did you install the rdkit from? -
• If you are not using conda: how did you install the RDKit? pip install