Add get_application_registry (plus prettify)

[Silmathoron]

This PR adresses #716 and helps solve #725.

I also moved the __repr__ to PrettyIPython so as not to shadow the pretty representation.

Because the doc uses the tuples with pickle, I also added a warning when making tuples, though I don't understand why you do not advocate for direct pickling...
Checks are made in UnitsContainer to avoid issues with multiple inheritance hierarchies in quantities and units

[cpascual]

maybe I am missing something, but... where is the registry variable defined?

[Silmathoron]

Obviously you're not and I was so focused on the pickle part that I did not even check the get_app_registry because it looked "so simple to do", and forgot to delete this line...

[Silmathoron]

I'm having trouble understanding the check that fails with travis... could you help me out?

EDIT: by the way, do you want me to add some Travis checks? Given what I just did before, it would probably be a good idea to check that the function executes correctly...

[iankharris]

While we're at it, why don't we update the docs to mention the pint.set_application_registry function? It's by far the easiest way pint to correctly unpickle pint units (in my opinion). It seems like such a trivial change that it's almost a shame to make a pull-request for this, but I really wish the documentation had alluded to this function before I trying to make the to_tuple method described in the docs to work with dask (it doesn't).

[Silmathoron]

Ok, I'll try to update the doc at some point.
It's going to take me some time to correct the pandas errors because of this... which I think should be changed to >= '0.24' anyway.

However, I still don't know why this call to data causes the pytest error on L967... and I don't think this has anything to do with my PR.

[Silmathoron]

Hey, there!
Now that pandas 0.24 is out, I thought I would check again what's going on, but it seems the notebook is not working with 0.24... I get some nasty accessors and delegated objects... can anyone tell me what's the status of the pandas support?

[hgrecco]

Closing this due to #880