MolToInchi adds spurious /b stereo for symmetric ylidene with 2D coordinates

Start by reproducing the supplied MolBlock and SMILES cases through the Chem.MolToInchi entry point, then trace the RDKit/InChI integration to determine where symmetry-equivalent paths gain the /b layer. The work is done when both representations produce equivalent stereochemistry-free InChI output and a regression check covers the example.

Describe the bug
Chem.MolToInchi() incorrectly infers double-bond stereochemistry for a symmetric ylidene molecule from its 2D coordinates, adding a  /b  layer even though the two substituent paths are symmetry-equivalent and E/Z stereochemistry is undefined. Doesn't happen when the molecule is provided as a SMILES.

To Reproduce

from rdkit import Chem, rdBase
from rdkit.Chem import CanonicalRankAtoms

molblock = """
     RDKit          2D

 12 12  0  0  0  0  0  0  0  0999 V2000
   -4.8794    0.4883    0.0000 C   0  0  0  0  0  0  0  0  0  0  0  0
   -3.7700   -0.5213    0.0000 O   0  0  0  0  0  0  0  0  0  0  0  0
   -2.3410   -0.0652    0.0000 C   0  0  0  0  0  0  0  0  0  0  0  0
   -2.0214    1.4003    0.0000 O   0  0  0  0  0  0  0  0  0  0  0  0
   -1.2315   -1.0748    0.0000 C   0  0  0  0  0  0  0  0  0  0  0  0  0
    0.1975   -0.6188    0.0000 C   0  0  0  0  0  0  0  0  0  0  0  0
    1.3069   -1.6283    0.0000 C   0  0  0  0  0  0  0  0  0  0  0  0
    2.7359   -1.1723    0.0000 C   0  0  0  0  0  0  0  0  0  0  0  0
    3.0555    0.2932    0.0000 C   0  0  0  0  0  0  0  0  0  0  0  0
    4.4845    0.7493    0.0000 C   0  0  0  0  0  0  0  0  0  0  0  0
    1.9460    1.3028    0.0000 C   0  0  0  0  0  0  0  0  0  0  0  0
    0.5170    0.8468    0.0000 C   0  0  0  0  0  0  0  0  0  0  0  0
  1  2  1  0
  2  3  1  0
  3  4  2  0
  3  5  1  0
  5  6  2  0
  6  7  1  0
  7  8  1  0
  8  9  1  0
  9 10  1  0
  9 11  1  0
 11 12  1  0
 12  6  1  0
M  END
"""

mol = Chem.MolFromMolBlock(molblock)
assert mol is not None

ranks = CanonicalRankAtoms(mol, breakTies=False)
assert ranks[6] == ranks[11]

molblock_inchi = Chem.MolToInchi(mol)
print(f"MolBlock InChI: {molblock_inchi}")
assert "/b" in molblock_inchi

smiles = Chem.MolToSmiles(mol)
smiles_mol = Chem.MolFromSmiles(smiles)
assert smiles_mol is not None

smiles_inchi = Chem.MolToInchi(smiles_mol)
print(f"SMILES: {smiles}")
print(f"SMILES InChI: {smiles_inchi}")
assert "/b" not in smiles_inchi

MolBlock InChI: InChI=1S/C10H16O2/c1-8-3-5-9(6-4-8)7-10(11)12-2/h7-8H,3-6H2,1-2H3/b9-7-
SMILES: COC(=O)C=C1CCC(C)CC1
SMILES InChI: InChI=1S/C10H16O2/c1-8-3-5-9(6-4-8)7-10(11)12-2/h7-8H,3-6H2,1-2H3

Expected behavior
The MolBlock- and SMILES-based representations should produce equivalent stereochemistry-free InChI output without a  /b  double-bond stereochemistry layer.

Screenshots

Configuration (please complete the following information):
• RDKit version: 2026.09.1pre (source commit  1df4988c99737db4e8ca722a5471526dd0271fbd ) -- but the issue is on all commits that have inchi support, InChI upstream misinterpretation
• OS: Amazon Linux 2023.10.20260302, aarch64
• Python version (if relevant): 3.11.14
• Are you using conda? No.
• If you are using conda, which channel did you install the rdkit from? Not applicable.
• If you are not using conda: how did you install the RDKit? Manual build from source.