Update Doros for new entries

[JoschD]

The Doros team added more entries on the root-level of the HDF5 files.
Before, there were only BPMs and METADATA and the BPMs were identified by not being METADATA.
As now there is also TIMESTAMPS_INDEX and TIMESTAMPS_TABLE (I told Manuel this maybe should go into METADATA, as METADATA itself seems to be empty anyway... but so far they have not moved it).

I considered going by name (e.g. find _DOROS in the name) but then we are limited to which machine data we can write into this format.
I now identify BPMs as entries that have a subgroup nbOrbitSamplesRead which should be pretty robust.

Added a test with a reduced version of the new data (script to generate the data is also attached).

[JoschD]

PS: with regards to the h5repack: This seems to be the only way to actually delete data from a hdf5 file (https://stackoverflow.com/questions/1124994/removing-data-from-a-hdf5-file). One should be aware of this. Quite annoying...

[fsoubelet]

Regarding the h5repack thing: how about we just create a new HDF5 file from scratch, and write into it the relevant parts? This way without copying first we don't have to worry about data deletion in HDF5. What do you think?

[JoschD]

Regarding the h5repack thing: how about we just create a new HDF5 file from scratch, and write into it the relevant parts? This way without copying first we don't have to worry about data deletion in HDF5. What do you think?

I was thinking about it, and even trying it ... but then you kind of need to know what the other data is, that is in the file and how many "levels" deep it is in the hdf5, as you cannot simply copy a whole "tree" into another file. And if we build up a fake-file from scratch, we might miss something that they (the DOROS people) put in that make the reader fail (as was the case with the new entries leading to this PR).
This script is only used in case we ever need to update the test-file anyway. I don't care if it crashes or is not the most optimal. I wanted a quick way to reduce the 20-30MB DOROS files while keeping their internal structure as intact as possible.

[fsoubelet]

In this case I'm all ok with this PR, thanks!