Function grouped_mat_mul_shared_owned 

pub fn grouped_mat_mul_shared_owned<T: MatrixScalar + TensorScalar>(
    lhs: Vec<T>,
    rhs: Vec<T>,
    output: Vec<T>,
    jobs: &[GroupedGemmJob],
    options: GroupedGemmOptions,
) -> Result<Vec<T>, GroupedGemmError>

Execute grouped GEMMs while consuming all three flat buffers.

The returned vector is the supplied output buffer after the grouped operation. Consuming the inputs avoids caller-side ownership bookkeeping; it does not duplicate their payloads inside the grouped descriptor bridge.

Errors

Returns GroupedGemmError using the same validation and provider rules as grouped_mat_mul_shared.