rename and update LHDFStore + add open_hdf (like in xw_excel.py)

[alixdamman]

TODO:

• add open_hdf function
• implement put(key, value, attrs) method:

   arr = ndtest((3, 3))
   with LHDFStore(filepath) as store:
       store.put(key='arr', value=arr, attrs={'type': 'Array'})

• update documentation of argument filepath of Axis/Group/LArray.to_hdf to tell it can also be a LHDFStore instance

[alixdamman]

@gdementen feel free to update to update my first comment.

[gdementen]

add open_hdf function

We could also go the other way round and deprecate open_excel. I am unsure about this. I never liked open_excel but maybe this is more intuitive for users??? We should make a poll :)

[gdementen]

The main reason for this issue is that I want to handle the case where users load or save an HDF file object by object (for whatever reason) without opening and closing the file for each object (like what happens when you use to_hdf/read_hdf). Maybe renaming LHDFStore and making it into a Session subclass (eg HDFSession) is the right thing to do. In that case, we will need to also make open_excel return an ExcelSession too. It seems like a good idea now, but this is probably hard to achieve because excel.Sheet objects are not really LArrays. In fact, I think they should be (or some subclass of them like LazyLArray), and their .data field should point to the sheet range. Probably a great thing but huge refactor would be necessary to achieve that.

[alixdamman]

See PR #761