Well, just tried looking through the MaxScript documentation, but it's practically useless with zero information, the same way I remember it from years ago, so no idea how to print a struct.
I got the 3ds Max 2016 trial, and I am looking into it.
There are two separate issues.
The first is what I mentioned before, where the exporter assumes the KEVT chunk is optional (something I actually asked BlinkBoy in the past about).
Since the model can't be loaded by the World Editor, this is wrong (yey, I was right all along).
The second problem, and the reason you encountered the first one, is that the script does not keep the keys correctly.
The reading code reads everything properly, the writing code writes everything properly (albeit I have no idea why, MaxScript looks weird!).
It's just that the keys are not there.
The next step is to figure why the keys disappeared.
-----
Ok, problem solved.
Finding it would have taken much less time had I not forgotten to add this to the Mdx sanity test.
Some of the event object tracks do not exist in any sequence, and are thus useless.
NeoDex removes those (regardless if you choose to optimize or not, this is done at import time).
Then it goes to problem 1 - missing the KEVT chunk, since some event objects now have no keys.
See attached file for the fix (event objects with no keys are not written in the first place).
-----
Whoops, this is not the correct fix after all.
The model isn't corrupted, but behavior is not the same (e.g. guts show when they shouldn't).
This means that our assumption of what tracks are considered useless is wrong.
-----
So after wasting all of this time, I am not going to fix this.
The fix is getting too involved with the rest of the code (mostly because sequence intervals get transformed for some reason I don't understand), and really should be done by BlinkBoy.
Sorry.
If you do want to use that model, you can open it with Magos first, and move the keys yourself so that they exist in actual sequences (but will this result in actually correct behavior? I don't know, this needs to be tested).
I attached the relevant sanity test result.