[FEA] pipe #239

Is your feature request related to a problem? Please describe.

It'd be super convenient to have a graph-friendly pipeline operator, replacing a(b(c(g))) with g.pipe(c).pipe(b).pipe(a), similar to pandas and other ~monadic envs

Describe the solution you'd like

It should support transforms over nodes, edges, and graphs

Ex:

g.pipe(graph=lambda g: g.bind(x='x'))
g.pipe(lambda g: g.bind(x='x')) # shorthand

g.pipe(edges=lambda g: g._edges[['src', 'dst']]) # g in, df out
g.edges(lambda g: g._edges[['src', 'dst']]) # shorthand

g.pipe(nodes=lambda g: g._nodes[['node']]) # g in, df out
g.nodes(lambda g: g._nodes[['node']]) # shorthand

• It should be allowed to provide both nodes=, edges=
• If both nodes= / edges= and graph= kwargs are provided, run them in the provided order:

g.pipe(nodes=fn_1, edges=fn_2, graph=fn_3)
g.pipe(graph=fn_1, nodes=fn_2, edges=fn_3)

Describe alternatives you've considered
While we do have g.edges(df) / g.nodes(df), they do not support flows like g.cypher(...).edges(clean_fn)

Additional context

• Similar to pandas pipe
• Most composition operators get inherited from here as they're table level
• ... Except we still don't have graph-level composition: union, subtract, ...