Function classify_op 

pub fn classify_op(
    actor: OpActor,
    op_kind: &str,
    op_address: &Address,
    op_index: Option<usize>,
    scope: &TurnScope,
) -> AffectednessClass

Classify a single operation against the current turn scope.

affects(O,S) = conflict(O.target, read_set) ∨ conflict(O.target, write_set). An op that intersects no part of the scope is Independent. Within the scope, a live_buffer actor is ProvenanceSpoofed (a lagging sidecar, not a real edit), a remove/move is StructuralDependency, a read-set intersection is InputAffecting, and a write-set-only intersection is OutputContended.

op_index is the op’s within-component node index (after-index for inserts/replaces, before-index for removes); it narrows exchange affectedness to the active tail (#loop-guard-exchange-node-granularity).