Check that cythonisation of XDR reader didn't introduce C int overflow bug similar to that in #4039

[hmacdope]

Cythonisation of the DCD reader in #3888 resulted in an int overflow bug as detailed in #4039 and fixed in #4084.

Given I did something similar for the XDR reader in #3892, we should check that no similar bugs were introduced.

[IAlibay]

I've tested this in two ways:

1. Create a ~ 16 GB XTC trajectory of 100000 frames as per Error when loading multiple large DCD trajectories #4039 using the TPR system.
2. Create a small 10000000 frames (~ 1.2 GB) XTC trajectory for a subset of the TPR system.

And then trying to load both to memory. In both cases I didn't encounter any issues.

[hmacdope]

@ianmkenney to test quickly on a large XTC privately and then good to go

[ianmkenney]

I iterated over a TRR file:

04:18 PM >>> du -sh production.trr 
292G    production.trr

and I didn't get a segfault

In [3]: u = mda.Universe("top.dms", "production.trr")

In [4]: from tqdm import tqdm

In [5]: for ts in tqdm(u.trajectory):
   ...:     pass
   ...: 
100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 191666/191666 [45:15<00:00, 70.58it/s]

In [6]: mda.__version__
Out[6]: '2.5.0-dev0'

In [7]: u.atoms
Out[7]: <AtomGroup with 136294 atoms>

Hope that helps!

[hmacdope]

With that I think we can close. Thanks @ianmkenney!