Add dataframe groupby/aggregate + graph reindex operators

[slabasan]

• Add dataframe groupby/aggregate operator
• Add graph reindex operator
• Add tests

[slabasan]

Should node be appended as an index level if a user wants to groupby module for example? The current graphframe implementation requires that node be an index level.

[slabasan]

If I want to groupby module, but the node has None for the module, do we drop this node?

[slabasan]

I'm trying to address this test graph, but I'm the connections aren't being set correctly. Assuming I want to groupby-aggregate by modules, I have node B and F which should be grouped in the same super node.

Original graph and associated dataframe:

0.000 A
├─ 5.000 B
│  └─ 5.000 C
│     └─ 1.000 D
└─ 10.000 E
   └─ 1.000 F

              name  time (inc)  time module
node                                       
{'name': 'A'}    A       130.0   0.0   main
{'name': 'B'}    B        20.0   5.0    foo
{'name': 'C'}    C         5.0   5.0   graz
{'name': 'D'}    D         8.0   1.0   graz
{'name': 'E'}    E        55.0  10.0    bar
{'name': 'F'}    F         1.0   1.0    foo

I want:

0.000 main
├─ 5.000 foo
│  └─ 5.000 graz
└─ 10.000 bar
   └─ 5.000 foo (same foo as above)

[slabasan]

This is ready for review.