lts.override.aut — sanctioned per-model LTS override (user-authorized).

Proven equivalent modulo the injected ports: verify PASS (verdicts identical
to dzn) AND wf-check EQUAL after hiding the injected-mechanism labels (the
open `c.h <=> *` and the `requires injected ihello w` ports) on both sides.
The only lts divergence is labeling: nucleos names the injected calls by the
CALLER port (sut.left.w / sut.right.w); dzn by the CALLEE (sut.c.h). modulo
the injected ports, ours == dzn. Hold nucleos lts (des (0,80,68)) as golden per
no-overfit-to-dzn. Was lts.xfail (incomparable). dzn gold kept as lts.aut.

---------- prior lts.xfail note ----------
# Expected to fail in lts category — known regression.
#
# This model fails `nucleos lts` weak-failures comparison against
# the dzn-generated `baseline/lts.aut`. The failure has been
# present long enough to be characterised:
#   - cluster of foreign-port / injected-port / fork / requires
#     models that exercise similar codepaths
#   - documented in memory [Parallel BFS bug] +
#     [Bail audit 2026-05-13]
#
# nuctest treats this as XFAIL (does NOT count toward CI exit-1)
# while a fix is pending. When the underlying bug is fixed and
# this model starts passing, nuctest will emit UNEXPECTED PASS
# in the summary header — that is the cleanup signal: delete
# this marker in the same commit as the fix.
