fix(KIM): use local cyclotron detuning values

[marjohma]

Closes #272

Summary

• build boundary harmonic detuning from the supplied plasma object instead of the module-global object
• use cell-center cyclotron frequency in the cell-center resonance argument
• extend the direct background regression across m_phi = -1, 0, +1 with distinct local/global and boundary/center values
• include no documentation files

Red/green evidence

Before the source fix, the new test failed on the boundary path because x2 used the module-global omega_c (got -18396.4; expected -18594.4). After the two substitutions it passes both boundary and cell-center checks.

Verification

• test_collision_scale_z0
• test_ion_temperature_gradient_model
• test_periodic_background_build
• test_flr2_fourier_kernel

All four pass locally after the commit. Independent review found no correctness issues.