slowrot: time/frequency-dependent detector-response likelihoods (Path…

[oshaughnessy-junior]

… A/B rotation + Path D finite-size)

Squashed feature import onto rift_O4d (dev history on rift_slowrot / calmarg_in_loop). Adds RIFT/likelihood/{factored_likelihood_with_rotation,factored_likelihood_freqresponse, slowrot_response,slowrot_freqresponse}.py, test_slowrot_*.py, and demo/rift/slowrot/. Wires --rotation-slow/--rotation-p-max (Path A/B) and --freqresponse (Path D) into integrate_likelihood_extrinsic_batchmode (CPU vectorized/NoLoop, guarded, opt-in). Validated: rotation NoLoop==baseline 3.6e-12; finite-size L->0==baseline 3e-9, Cauchy-Schwarz respected, V4 positive control +38.9 nats; cubic time-interp with precision-preserving time ref.

[oshaughnessy-junior]

Auto-review at head 6d8fa047c9aacb9d752eab4e152a285aa797d03b against current rift_O4d 85b451dc1d4c3f9788491e1dfb613a68b5a2d7ec:

• This PR is superseded by its target branch. The stable patch ID for the sole PR commit is 0226b1a725a1c8eb61035683507900235908dcf6, exactly matching target commit 4d6e66f09cc463ba93b5c2f45368c66c2642daae. Current rift_O4d also contains later slowrot GPU/CuPy/ILE validation fixes, so replaying this older squashed patch would risk regression.
• GitHub reports CONFLICTING/DIRTY; git merge-tree --write-tree confirms conflicts in factored_likelihood_freqresponse.py, factored_likelihood_with_rotation.py, integrate_likelihood_extrinsic_batchmode, and demo/rift/slowrot/README.md.
• Clean-head checks: git diff --check passed; all touched Python files compiled; the focused suite excluding one compatibility-stopped module reported 28 passed, 4 warnings. Full collection stops on local Python 3.13 / NumPy 2.4.6 because test_slowrot_noloop_bruteforce.py still calls removed np.trapz twice. The current target has the same calls. With a diagnostic np.trapz = np.trapezoid alias, the numerical comparison passes with max timeseries difference 3.160e-10.
• The LDAS anchor ~/RIFT_roboto_paper/analyses/slowrot_demo/ is present and contains the saved quick-look gain results; its full injection-recovery posterior-bias / sky-map DAG remains TODO.
• SLOWROT_HANDOFF.md needs a readiness cleanup on the active branch: the top says Path B is production-ready for physical targets at p_max<=2, while the stale lower status section still says not to ship p>=1.

Recommendation: close #157 as superseded instead of resolving it. On active rift_O4d, replace or compat-wrap np.trapz, reconcile the Path B readiness text, and keep the full LDAS DAG as the remaining external science gate.