Remove redundant "tail" argument from memmap loading

[oerc0122]

No description provided.

[abuts]

This change is incorrect and harmful.
As soon as pixels are last dataset in sqw file, all this works. As soon as you write something after pixels when updating some datasets, it fails with very strange error if tail is not present.

It is of course very bad that unit tests do not cach this change, though I believe I was writing some tests for that.
This ticket may be modified to write unit tests which check tail presence necessary

[oerc0122]

This change is incorrect and harmful. As soon as pixels are last dataset in sqw file, all this works. As soon as you write something after pixels when updating some datasets, it fails with very strange error if tail is not present.

It is of course very bad that unit tests do not cach this change, though I believe I was writing some tests for that. This ticket may be modified to write unit tests which check tail presence necessary

That sounds like it's masking some other error and we should definitely not have this. Retry since I've rewritten the PixelData dumpers.

[abuts]

I've observed failure when tail is not there more then once. You've already tried to remove tail, and I have to restore it as tests started to fail randomly. The test, which writes something after pixels and shows that you can still read pixels without tail is absolutely necessary to prove you can remove it. It is test for memmapfile rather then for anything else

[oerc0122]

And thus it is clearly masking a deeper problem. This is not a solution.