- add a new parameter in settings 'H5_ZSERIE_INIT_TITLES'
  will be merge with 'H5_INIT_TITLES' when h5_to_nx is call
- split _Acquisition into a base class (_BaseAcquisition and _ZSerieAcquisition). PR of ID15 data will need to be rework due to this I guess.
- _ZSerieAcquisition is a serie of _StandardAcquisition. 'write_as_nxtomo' is now only part of StandardAcquisition

- now we will first try to get the camera using NX_class == 'NXdetector' from the instrument group
- if fails then we try to retrieve group that looks like detector (containing `data` dataset which should be 3D)
- users can still force camera names. Either from settings.py of from command line parameter '--valid_camera_names'

- if true then raises ValueError if cannot reach some requested data
- if false the set a default value and add a warning message to logs

Conflicts:
	nxtomomill/settings.py

- only write pixel size. A path can still be provided by the user. Otherwise we will try the following path to get the pixel size to be store in the NXTomo (with this order):

"technique/optic/sample_pixel_size ",
"technique/optic/sample_pixel_size",
"technique/detector/pixel_size"

/close #20

[converter][h52nx] Instead of defining the scan titles in AcquisitionStep use a dedicated structure 'H5ScanTitles'