HK evaluation bug fix

The HK probability density function is a semi-infinite integral that was evaluated until recently from 0 to np.inf. However, after some testing, I have changed that in 3e52d03 from 0 to 500 because it was increasing the computation time a little bit. However, in some extreme cases with very high coherent content, it appears to be not enough. So I reverted back to the initial settings.