fix(KIM): restore collisionless j-B parity

[krystophny]

Summary

• evaluate the collisionless periodic magnetic-current factor with the derived regularized combination z0/kAbs
• preserve the magnitude-based stable response argument and the causal pole used by the magnetic-charge block
• add a direct positive/negative-k_parallel parity regression and update the corrected manufactured current reference

Verification

• red: the new parity assertion failed against the former z0/kPole implementation
• green: ctest --test-dir build -R '^(test_collisionless_fourier_kernel|test_collisionless_ion_assembly|test_periodic_assembly|test_periodic_solve|test_kim_solver_periodic)$' --output-on-failure
• result: 5/5 tests passed
• independent review: no critical findings; the bounded numerical correction is consistent with the signed derivation

Documentation follow-up

The external consolidated derivation and Mathematica source still encode the old recurrence and therefore do not regenerate this corrected oracle. That out-of-repository follow-up is recorded explicitly in #263; this PR does not modify external documentation.

Closes #254
Related: #263

[marjohma]

Closing this draft so it can be recreated under the requested marjohma PR authorship. The replacement will use the same branch and commit.

[marjohma]

Replacement: #278, opened by marjohma from the same branch and commit.