Describe the bug
Tautomer canonicalization ends up clearing/losing stereo in this case

To Reproduce

from rdkit import Chem
from rdkit.Chem.MolStandardize import rdMolStandardize

rdkit_mol = Chem.MolFromSmiles("O=C(N1)[C@H](CC(N)=O)CC1=O")
assert Chem.MolToSmiles(rdkit_mol) == "NC(=O)C[C@@H]1CC(=O)NC1=O"
enumerator = rdMolStandardize.TautomerEnumerator()
enumerator.SetMaxTautomers(10000)
enumerator.SetMaxTransforms(10000)
enumerator.SetRemoveSp3Stereo(False)  # default=True
canon_mol = enumerator.Canonicalize(rdkit_mol)
assert Chem.MolToSmiles(canon_mol) == "NC(=O)Cc1cc(O)[nH]c1O"  # but it should not have lost it

Expected behavior
To get "the most canonical" tautomer without losing that bond. Or have some flag to make it happen. In OE toolkits, these transformations can be disallowed.

Configuration (please complete the following information):

• RDKit version: 2024.9.2
• OS: Ubuntu 20.04.2
• Python version (if relevant): 3.9
• Are you using conda? No
• If you are using conda, which channel did you install the rdkit from? n/a
• If you are not using conda: how did you install the RDKit? Built from source