[mini] Fix the bug in Issue #377: Add original duration into the argument of get_phi2

[huixingjian]

When introducing STC, the laser duration may be changed by both spatial and angular chirp.

The spatial chirp $\zeta$ will stretch a Gaussian pulse as $\tau' = (\tau_0^2+\frac{4\zeta^2}{w_0^2})^{1/2}$.

The temporal chirp, i.e GDD, will stretch a Gaussian pulse as $\tau= [\tau'^2+\frac{4(\varphi^{(2)})^2}{\tau'^2}]^{1/2}$.

Previously, the function get_phi2 assumes that $\tau'\approx \tau_0$. It's true when the pulse is based on a Gaussian shape longitudinally(The influence on Gaussian laser duration from spatial chirp is always minor). However, as pointed out in Issue #377, when implement PolynomialSpectralPhase on the laser, the pulse may be drastically stretched by both spatial and temporal chirps. This causes an error.

I tried to calculate the GDD in spectrum field as $\langle \frac{\partial^2 \varphi}{\partial \omega^2} \rangle$, but in this way turns out to be really sensitive to the box size and resolution, which originated from two times of derivative in the same direction plus numerical FFT. Considering $\tau'$ is also not a trivial to calculate, currently the solution is to add one argument to input the original value of $\tau_0$ for user to use.

The final optimal solution remains to be discussed...