Avoid duplicated repos.

[frankfliu]

This fixes #80.

This might not be the best solution, it assumes DependencyResolver is an instanceof IBiblioResolver (this is not future proof).

[SpencerPark]

@frankfliu I like this, but maybe instead of looking for duplicate urls, we could skip duplicates based on the name.

A name should uniquely identify a resolver and that property is exposed on DependencyResolver. This ofc doesn't avoid the case where the same repo is added under multiple names but I don't think that is a case we should worry about. I believe that should however cover the common case outlined in #80. What do you think?

[frankfliu]

I still have concern regarding name conflict. One of the use case is I opened multiple notebook, they are actually shared the same JVM, they actually can see each others resolver (I'm not sure if this is a bug). In this case, there are a high chance two notebook using the same common name like "s3" or "maven" for different URL.