Initial GaaS working version

[rlratzel]

DGL is the initial use case this is supporting.

---

• Create client class
• Create Cluster (Need ability to start dask cluster for MG APIs #3)

Load Data / Create Graph

• loading graph data from a file on the machine the service is running from
• unloading/removing the loaded graph
• research and implement ETL functionality (Add ability to execute arbitrary data loading and ETL operations on graph input data #4)
• Add support for MG property graph (Add support for using an MG Graph/PropertyGraph #5)

Run Sampling

• Running and returning results for node2vec sampling on the loaded graph. Results are returned to caller's host for now.
• Running and returning results for egonet sampling on the loaded graph. Results are returned to caller's host for now.
• Running and returning results for neighborhood sampling on the loaded graph. Results are returned to caller's host for now (Add Uniform Neighbor Sampling API #6)