v0.442.0 — Q-251 PRE-STATED TEST SHIP

B288 (PAPER_2283): the first fully pre-registered test in the corpus, a
second liquid, and neither prediction survived. One band since v0.441.0.

- THE PRE-STATEMENT: written into RULINGS_QUEUE.md Q-251 and committed to
  the repository before any run. Lennard-Jones argon at the NIST 85 K /
  0.101325 MPa state point (c_0 854.35 m/s, eta 0.2795 mPa s, read from
  the WebBook). P1, the Q-247 form: k_c = omega_SCm/c_0 = 9.19 nm^-1. P2,
  the Q-251 route: (D_BSFG/D_phys) x P1 = 13.79. Rule: +-10 pct. The
  result is appended beneath the pre-statement, which is unchanged.
- THE INSTRUMENT: the B287 engine's LJ path (md_grade/lj_kernel.c,
  argon_record.py, gridcur.currents3). Kernel exact against numpy; forces
  = -grad E to 6e-9; NVE +0.00001 kJ/mol per atom over 10 ps. The model
  reproduces NIST: eta_0 0.274 / 0.277 vs 0.2795; D 1.8e-5.
- THE MEASUREMENT: 4096 atoms, 2 seeds x 400 ps, 36 shells to 26 nm^-1.
  eta(k) Gaussian over the water-comparable range (chi2 43/46 vs
  Lorentzian 94/111): k_c = 7.25 / 7.09 -> 7.17 +- 0.05 nm^-1.
- THE RULE APPLIED: k_c/P1 = 0.78; k_c/P2 = 0.52. NEITHER. Q-251 CLOSED
  NEGATIVE - water's 1.489 was a coincidence. P1, already negative for
  water (+49 pct), is negative for argon (-22 pct).
- THE OBSERVATION (recorded, not claimed): sound speeds differ 1.73x, the
  cutoffs 1.10x; k_c x sigma = 2.50 (water) and 2.44 (argon). The eta(k)
  cutoff tracks the molecular diameter, not omega_SCm/c_s. Q-252 OPENED,
  Daniel-gated: a category error (Theorem B untested by front 4), or the
  framework fluid cutoff falsified as written.

Not claimed: which reading of Q-252; the model's own c_0 (not cleanly
measured; P1 evaluated with NIST's as pre-stated); the fit range (not
pre-specified; both reported; neither rescues a prediction); anything
beyond two liquids at one state point each.

Gate 6,115/0. Dispatch 2,334 keys. Registry 7,141 rows. Acceptance
109/109 (simulator untouched).
