Issues with Hillas parameters

There are several issues with Hillas parameters. When analysing gamma_test_large.simtel.gz from ctapipe-extra, I get some strange results:
(you can see notebook hillas_dataframe to test this)

• width > length ~25% of the time
• phi and psi show large values (-50 to 400) while the function telescope_reco.get_hillas_parameters_container states that they are in radians
• -pi/2 < phi < pi/2 instead of [-pi:pi]
• Kurtosis very different from ctapipe
• hipecta and ctapipe psi don't have the same reference
• skweness very different from ctapipe
• width very different from ctapipe