Function emit_plan_stim 

pub fn emit_plan_stim(
    plan: &NodeEmissionPlan,
    options: &PlanStimOptions<'_>,
) -> Result<PlanStim, StimEmissionError>

Lower one node’s emission plan to standalone Stim text.

Every Op variant is handled, so a plan is never rejected for containing an operation this does not know about. Two conventions are worth naming because they are not the only defensible choices:

• REPEAT is unrolled. Self::measurement_columns are absolute positions, and a measurement inside a loop occupies a different one every iteration. Unrolling is what makes the map well defined; the cost is that a plan with a large repetition count produces proportionally large text.
• A feedback Y is written as CX then CZ, not CY. The control is a classical record, so the phase difference between Y and XZ is constant across the whole branch and cannot be observed. This matches the convention the artifacts already in the field were built with, and deliberately differs from emit_bloq_stim, whose text is pinned by its own consumers.

Errors

• StimEmissionError::Circuit if a plan qubit is missing from the qubit map, a body reference dangles or recurses, or a record lookback leaves Stim’s range.
• StimEmissionError::QubitIndexOutOfRange if a caller-supplied qubit index exceeds 16_777_215.
• StimEmissionError::UnsupportedGate for a non-Clifford gate other than T/T_DAG in the StimDialect::Stim dialect, which Stim could not parse.
• StimEmissionError::MalformedGraph if a feedback control is read before its measurement is emitted, or the plan names a measurement its own circuit never emits.

Panics

In debug builds, panics if plan emission disagrees with the circuit column walk.