Oliver Beckstein

Super-quick comment: Looks as if you addressed all concerns. I'd merge correct code even if there's still optimization to be done. However, is there any performance regression for the cases...

Checking that `u = mda.Universe(tpr_file, tpr_file)` and `u = mda.Universe(tpr_file)` both work would be good.

Yes, add a one off test, maybe it breaks in interesting ways in the future and reveals a real issue. Might eventually be useful when we optimize towards reading only...

Just after reviewing I looked at the docs and realized that the new TPR coordinate docs were not linked. I quickly added the files... hopefully all looks good.

Congratulations @tylerjereddy for closing a decade-old issue! (Btw, I carefully preserved and edited your log messages in the squash-commit as they contain a plethora of useful information.)

@richardjgowers could you please have a look at this PR, looks like your wheelhouse. If you don't have the bandwidth please un-assign yourself and let me know. Thanks!

@richardjgowers do you have the bandwidth to look after this PR? If not please unassign yourself. Thanks!

Sorry, I won't be reviewing, I haven't had to deal with pdbx & friends (and I have plenty of other PRs to review). Someone with experience should have a look....

@richardjgowers , were your earlier comments addressed and if so, would you approve this PR?

@hmacdope @richardjgowers are you still able to shepherd/merge this PR? Otherwise please unassign yourselves. Thanks!