Backprop: Single Linear Layer

Implement the backward pass for a single linear layer y = xW + b.

Signature: def backprop_single_layer(x: np.ndarray, W: np.ndarray, delta: np.ndarray) -> tuple

Return (dW, db, dx) where:

• dW = x.T @ delta — gradient w.r.t. weights
• db = delta.sum(axis=0) — gradient w.r.t. bias
• dx = delta @ W.T — gradient w.r.t. input