softplus and inverse_softplus warn about overflow when they should not

[han-ol]

These functions lead to a warning when expm1/exp overflows, but this overflow is already disarmed by np.where.

 def inverse_softplus(x: np.ndarray, beta: float = 1.0, threshold: float = 20.0) -> np.ndarray:
    """Numerically stabilized inverse softplus function."""
    return np.where(beta * x > threshold, x, np.log(beta * np.expm1(x)) / beta)

...

def softplus(x: np.ndarray, beta: float = 1.0, threshold: float = 20.0) -> np.ndarray:
    """Numerically stabilized softplus function."""
    return np.where(beta * x > threshold, x, np.log1p(np.exp(beta * x)) / beta)

The expected behavior is: Don't warn when such a caught overflow occurs.