Select molecular states and record preparation decisions for docking

[dprada]

What — Make the chosen ligand and receptor states and preparation decisions explicit in a docking run.
How — Record ligand chemical state and conformer, receptor state, pH and residue variants, retained waters/cofactors, and any transformation parameters; associate each result with those inputs.
Why — The Core MVP requires prepared states and backend representations to be traceable to source systems.
Record — devguide/pending_proposals/select_molecular_states_and_record_preparation_decisions_for_docking.md
Related — #3. DockingMT consumes MolSysMT state selection and preparation capabilities tracked in uibcdf/molsysmt#217, uibcdf/molsysmt#218, uibcdf/molsysmt#220, uibcdf/molsysmt#221, and uibcdf/molsysmt#222.

[dprada]

Partial progress in c68cc7d: DockingProblem already records source form, selected atoms and structure/chemical-state identifiers, conversion report, and file fingerprint. Vina preparation now explicitly records the polar-hydrogen policy, omitted hydrogen indices, and rigid torsion policy. Result provenance includes SHA-256 digests of the exact receptor and partner PDBQT bytes submitted to Vina. Guard: tests/test_redocking.py::test_direct_molsysmt_input_reaches_vina.

This proposal remains open. Validated charge/typing methods, protonation and pH choices, residue variants, and explicit water/cofactor decisions still need provider support or a documented preparation workflow. The PDBQT digests identify bytes but do not retain generated artifacts. Local report: devguide/pending_proposals/select_molecular_states_and_record_preparation_decisions_for_docking.md (partial).

[dprada]

The file-backed replay and provenance slice is complete in #10 (c044de4, 1a2f7d0). A saved 181L result reconstructs the recorded molecular problem and Vina protocol, checks source/PDBQT hashes and atom selection, and reports a controlled second run. The measured exploratory baseline is devguide/validation/181l_redocking_exploratory.md.

This advances recorded-configuration reproducibility. Issue #4 stays open for explicit molecular-state and preparation-decision provenance; the 181L preparation remains provisional under #5.