[rohithsrinivaas]

Adds InnerReorganizationMaker, a Q-Chem flow for the intramolecular (inner-sphere) reorganization energy λ of an electron transfer between two states of a molecule (e.g. a molecule and its radical ion), using the four-point scheme of Nelsen et al. The outer-sphere (solvent) contribution depends on the environment (donor–acceptor distance, electrode, ε/ε∞) and is left to post-processing.

Flow

1. Optimize both states (opt_maker, default FrequencyOptFlatteningMaker; None uses the given geometries).
2. Four single points at the same level of theory: each state at its own and at the other state's optimized geometry, with scf_convergence = 8 (Q-Chem's single-point default of 5 is too loose for meV energy differences).
3. analyze_inner_reorganization → InnerReorganizationDocument: λ on both surfaces and their mean, vertical and adiabatic energies (and ΔG with frequencies), the RMSD between the geometries, a connectivity check (flags ET coupled to bond
   breaking), and an electronic-state check (max_spin_change, electronic_state_change): a state whose Mulliken spin moves by > 0.3 on an atom between its two geometries is not the same electronic state at both, and its unconstrained λ
   mixes two surfaces.

Constrained-DFT diabatic mode

When the transferred electron localizes differently at the two geometries, the unconstrained λ is meaningless. With cdft_fragment=(first, last) and cdft_fragment_states=((q₁, m₁), (q₂, m₂)), the single points become constrained DFT
calculations that keep each state's charge (and spin) on the fragment at both geometries (default partition: fragment-based Hirshfeld). By default the unconstrained points are also run, giving the true vertical/adiabatic energies and the
diabatic–adiabatic gaps.

InnerReorganizationMaker(
    cdft_fragment=(1, 10),                   # ethylene carbonate (Li is atom 11)
    cdft_fragment_states=((0, 1), (-1, 2)),  # EC in Li+-EC, EC•⁻ in the reduced complex
)   

To support this, QCInputGenerator gets cdft_constraints (pymatgen QCInput.cdft format) and fragments. Note: Q-Chem 6.2 silently ignores $cdft when gen_scfman = true (atomate2's default) — the energy equals the unconstrained one
and no CDFT iterations are printed (checked on He₂⁺). The generator therefore sets gen_scfman = false whenever constraints are given.

Also

• SinglePointMaker, ForceMaker and TransitionStateMaker now set valid task_types ("Single Point", "Force", "Transition State Geometry Optimization"), like OptMaker/FreqMaker. Without them the task document stored the job name (e.g.
  "single point") as task_type, which is not a valid emmet TaskType, so the whole task document could not be loaded back (resolving a reference to it raised a pydantic ValidationError).

Tests

• NH₃ → NH₃⁺ (ωB97M-V/def2-SVPD, vacuum; real Q-Chem 6.2 data): vertical IE 10.90 eV, adiabatic 10.17 eV, λ = 0.225 / 0.723 eV.
• Li⁺–EC → Li⁺–EC•⁻ (ωB97M-V/def2-SVPD, custom SMD): unconstrained, the reduced complex at the Li⁺–EC geometry puts the electron on Li (λ₂ = 0.33 eV, flagged by electronic_state_change); the diabatic mode gives λ = 2.165 / 1.200 eV, with
  diabatic–adiabatic gaps of 0.04–0.05 eV at each state's own minimum.
• Unit tests for the document, fragments, CDFT input set, constraint bookkeeping, spin check and maker validation.

The end-to-end CDFT test needs emmet to parse Q-Chem inputs with fragments (and a custom SMD solvent), materialsproject/emmet#1537; it is skipped with an emmet that cannot.