pecos-neo importance runner silently drops every injected noise gate except X/Y/Z

[ciaranra]

The importance-sampling runner discards any noise gate a channel injects that is not a single-qubit Pauli.

ImportanceRunner::execute_noise_gate matches only X, Y, and Z, with a bare catch-all that does nothing:

// exp/pecos-neo/src/sampling/importance_runner.rs:592-608
match gate.gate_type {
    GateType::X => { self.simulator.x(&qubits); }
    GateType::Y => { self.simulator.y(&qubits); }
    GateType::Z => { self.simulator.z(&qubits); }
    _ => {}
}

Anything else a noise channel emits -- RZ from coherent idle dephasing, H, S, or any
multi-qubit noise -- is dropped without a warning or error. The simulation runs to completion
and reports results that are quieter than the configured noise model, with nothing indicating
that noise was discarded.

This is unconditional in this runner: it does not depend on how the runner was constructed.

Impact: importance-sampled results are silently wrong for any noise model configured with
mechanisms outside single-qubit Paulis. Because importance sampling is used specifically to
estimate rare logical failures, understated noise is the failure mode that matters most.

Suggested direction: either execute the remaining gate types, or fail loudly when a noise
channel emits a gate this runner cannot represent. Silently continuing is the one option that
should be off the table.

Found while reviewing after-2q idle noise in exp/pecos-neo; not addressed there because it
is independent of that change.

[ciaranra]

Fixed on PR #420 (commit 696dbc550). Injected noise gates are now executed where the simulator supports them, rejected at configuration time with a message naming the configuring setter and the runner, or panic as a broken invariant — never silently skipped.

Treating this as a class rather than two instances turned up a third, more severe case: exp/pecos-neo/src/tool/simulation.rs discarded the configured NoiseResource before it reached the importance runner, so importance sampling through the high-level API ran with no noise model at all. The comment justifying it ("IS uses its own boosted noise") was inaccurate — ImportanceRunner::with_noise already existed and self.noise was live in the shot loop. That is also fixed.