KineticForces - Read the kinetic file's zeff dataset #526

The kinetic profile file can carry an optional zeff dataset, which the SLAYER layer and the resistive-layer overlap scan prefer to their own closure (PR #298). The NTV path does not read it: resolve_ntv_species always derives Zeff from quasi-neutrality with one impurity ([KineticForces] zimp, mimp). One run can therefore use two Zeff profiles: the file's in the tearing analysis and the closure's in the NTV torque.

Why this is more than a one-line change:

• In the NTV path Zeff comes with the pitch-angle factor zpitch and the impurity density from multi_ion_composition, and the collision frequencies use all three. Overriding Zeff alone would leave them inconsistent.
• zpitch has a pole at Zeff = zimp, so a supplied Zeff above zimp needs a defined behaviour.
• The impurity charge is set twice, as [KineticForces] zimp (integer) and [SLAYER] zimp (float). They should become one setting, or be documented as deliberately separate.

Done when:

• A rule is recorded for what a supplied Zeff means for zpitch, the impurity density and the collision frequencies.
• The NTV path uses the file's zeff under that rule and logs which source it used.
• A test covers a file with and without the dataset.
• docs/src/kinetic_forces.md states the precedence for both paths.

🤖 Generated with Claude Code