Update Node route, new Node classes, altered POST data handling

[reelmatt]

A few change highlights:

• Added 'update' route for Node info
• Changed the Workflow add_node method to update_or_add_node
• Added skeleton 'PivotNode' and 'JoinNode' classes for front-end to load
• Changed handling of POST data from form-data to raw JSON.
  • Should allow for more complex JSON requests than simple key: value pairs
• Node attribute changes
  • Still trying to figure out the right balance between class/instance attributes and which (sub)class to declare them in.
  • Biggest addition is an options dictionary. Idea is this can store Node parameters (e.g. 'delimiter' for the ReadCsvNode. The idea is these could be passed in to whatever pandas function needs them, and are also used to feed front-end for modifying behavior.
• Some style/documentation additions/modifications.

Added some new Postman tests to hit new functionality. There may still be some wonkiness and KeyErrors, but most of it should be working.

[reelmatt]

@reddigari, your method was exactly what I had in mind and is what I was attempting to do. I made the constant name DEFAULT_OPTIONS with the thinking we might want something like DEFAULT_STYLES as well (color, port shapes/kinds?). Would love to hear what others think in terms of this approach and how it might scale/adapt going forward.

One example that might guide some thinking: I realized storing options in a dictionary means we can pass them in as **kwargs to a pandas call. However, instead of the file option that was at the IONode level, the two concrete nodes now have custom filepath_or_buffer and path_or_buf options to mirror the pandas function. This seams simpler for node execution, but might require more custom code for file handling/parsing?

[reddigari]

@reelmatt Definitely agree that passing the options as **kwargs is a good goal, but not necessary if it ends up being a hassle. One strategy I've used when wrangling keyword argument dicts is to extract the problem ones, pass them as positional or explicit keyword arguments, and then pass the remaining dict as kwargs:

file_buf = options.pop("file")
arg_name = options.pop("invalid_arg")
pandas.whatever(file_buf, arg_name=arg_name, **options)