Add trunk to StreamObject

[Teschl]

This implements the trunk function. I (finally) got the trunk function working, more or less copying the Matlab version directly but modifying it to work in the python version.

This still needs work:

• The variable names are taken directly from the Matlab version. I'm not a big fan of these variable names, they should be more descriptive. (also, they don't conform to pep8 in some places, so a failing pipeline is expected)
• It also doesn't support the use of arguments A (GRIDobj with flow accumulation values) and a (node-attribute list). Should we include this functionality (in parts or all of it)?

@wkearn Are there variable names that should be kept in the current form, or can I go ahead and replace them by more descriptive names? I assume some of them are more or less standard for what they represent.

closes #113

[wkearn]

This implements the trunk function. I (finally) got the trunk function working, more or less copying the Matlab version directly but modifying it to work in the python version.

Looks good, @Teschl! I am curious if we could do something to avoid the sparse matrix manipulation, but it makes sense to have a working version based on the MATLAB implementation that we can experiment with in the future.

This still needs work:

* The variable names are taken directly from the Matlab version. I'm not a big fan of these variable names, they should be more descriptive. (also, they don't conform to pep8 in some places, so a failing pipeline is expected)

@wkearn Are there variable names that should be kept in the current form, or can I go ahead and replace them by more descriptive names? I assume some of them are more or less standard for what they represent.

I think we can replace these variable names with more descriptive ones. OUTLET is the only scientifically meaningful one, users won't see any of this anyway, and we might as well follow the standard. I might suggest OUTLET => outlets, II => max_neighbor and I => trunks. Feel free to use ones that make more sense to you if you can think of some.

* It also doesn't support the use of arguments `A` (GRIDobj with flow accumulation values) and  `a` (node-attribute list). Should we include this functionality (in parts or all of it)?

This shouldn't be too hard to add because we have StreamObject.ezgetnal. You should be able to do something like

def trunk(self, a=None):
  # ...
  if a is None:
    dds = self.downstream_distance()
  else:
    dds = self.ezgetnal(a)

ezgetnal will throw an error if a is not the right data type or size.

closes #113