fetch_adjacencies() API is awkward

[stuarteberg]

To avoid a huge edge table with lots of redundant info, it returns two dataframes, separating edge data from neuron properties. But the result is annoying to deal with.

For example, to merge the properties from the pre neurons onto the edge table, you have to do something like this:

neurons, edges = fetch_adjacencies(NC(somaLocation=NotNull), targets, properties=['zapbenchId'])
neurons_pre = neurons.rename(columns={c: f"{c}_pre" for c in neurons.columns})
edges = edges.merge(neurons_pre[['bodyId_pre', 'zapbenchId_pre']], 'left', on='bodyId_pre')

It would be better to return three dataframes: neurons_pre, neurons_post, and edges.

And maybe it would be worthwhile to offer an option to pre-merge the neuron properties directly onto the edge table, assuming the results aren't too huge.

---

Update: Upon reflection, I think I'm okay with just breaking the API and adding a new output.