Function estimateFrontdoor

estimateFrontdoor(
    graph: CausalGraph,
    treatment: string,
    outcome: string,
    data: number[][],
    nodeIndex: Map<string, number>,
    mediators: string[],
): { ate: number; se: number }
Frontdoor adjustment ATE.

For treatment X → mediator M → outcome Y (with no backdoor X↔Y): P(Y|do(X)) = Σ_m P(m|x) Σ_x' P(Y|x',m) P(x')

Linear approximation: β_M * β_X (product of coefficients)
Parameters
- graph: CausalGraph
- treatment: string
- outcome: string
- data: number[][]
- nodeIndex: Map<string, number>
- mediators: string[]
Returns { ate: number; se: number }
- Defined in pipeline/src/infer/effect-estimation.ts:134