[BUG] Discover converter does not appear to be reading the correct timestep

[ChiCheng45]

Description of the error
Running a temperature calculation on the Discover trajectories found in the MDANSE-Examples repo shows large differences between the temperatures calculated using the velocities from the MD output and with numerical derivatives. Here are the results with order = 0 (blue) and order = 5 (orange) with cubane (top) and sushi (bottom).

The velocities from the MD output are correct since they give a temperature of around 300 and 200 K for cubane and sushi respectively. The numerical derivatives are correct since they provide accurate results with other trajectories. Both Discover trajectories have a time step of 1 fs seems suspiciously like some default value.