memory reader assumes all trajectories start at time 0

[kain88-de]

Expected behaviour

when I have a dcd/xtc that is part of a long simulation the first frame in the file should have a different time than 0.

Actual behaviour

the memory reader always assumes that the first frame has time 0. The code uses this

        ts.time = self.ts.frame*self.dt

frame is the actual frame in the trajectory but not the current step in the simulation. For correct behavior we
have to add a offset. Adding a simple offset breaks a lot of unit-tests though so there seem to exist some subtle corner cases.

Currently version of MDAnalysis:

(run python -c "import MDAnalysis as mda; print(mda.__version__)")
mda-0.17.1-dev

[orbeckst]

@richardjgowers did you look at this issue in your recent MR reboot PR #2080?

[richardjgowers]

Not looked at it yet, we could even try and slurp the time and add it as an Auxiliary or something..

[IAlibay]

After discussion on discord, this has been moved to a milestone for 3.0 (unless there's any disagreement with this change)