Improved documentation about time stamped data

[mkoennecke]

Refs #503
I have updated the information on storing time stamped data in NeXus files. I find this now sufficiently documented. Some one else, please review and merge. It is now much better then it used to be.

[vasole]

To replace something expected to be a dataset by a group is in itself not a good idea.

Furthermore, we are replacing it by a group where everything is optional. It would have been much, much simpler for a program to test if at the same level as NXdetector/data one could find a time or timestamp or whatever name to find the additional information.

That would prevent breaking of any code.

A smarter solution, still going in the direction of what this PR implements, is to keep the dataset data as it is, and to create an NXlog inside the NXdetector where value is a link to NXdetector/data with all the required time information. You get your NXlog and nobody gets broken code.

[mkoennecke]

Armando, I am sorry to disappoint you but this has been discussed and decided upon at NIAC 2016. It just took embarrassingly long to update the documentation. And anyway, you cannot use the same code for treating streaming data then for more traditional data collection schemes. Except if you adapt the code and then you can also put in the check for NXlog.

[vasole]

I agree you cannot use the same code to analyze the data, but you can use existing code to visualize the data in one case but not in the other.

[FreddieAkeroyd]

Related to #503

[PeterC-DLS]

allows to use can

[benajamin]

some small language corrections.

[FreddieAkeroyd]

Looks good

[vasole]

@mkoennecke

Another resolution against common sense approved in a meeting where I was not present.

How many NIAC representatives are actually developing and supporting code?

This resolution is the equivalent of changing the signature of a function instead of adding an optional argument.