Add !mermaid command

[ryannedolan]

Added a new CLI command !mermaid which renders pipelines as mermaid flowcharts. In the resulting graph, resources are grouped by "kind".

Added a builder DSL to ResourceProvider to enable expressing source->sink relationships between resources. These relationships are not meaningful to the planner or operator.

Testing

The existing adapters are too simple to adequately test this against, so I modified the Kafka adapter to include additional "hops" (not included in the PR):

ResourceProvider.empty()
  .with(x -> new MySqlTable(x))
  .to(x -> new BrooklinCDC(x))
  .to(x -> new KafkaTopic(x, numPartitions, topicConfigProvider.config(x)))

Using this somewhat contrived adapter, I tested a simple query across two tables:

0: hoptimator> !mermaid SELECT * FROM RAWKAFKA."products", RAWKAFKA."test-topic"

... which produces the following flowchart:

flowchart
  subgraph SqlJob
  R663951423["sql: CREATE DATABASE I..."]
  end
  subgraph BrooklinCDC
  R1206753277["name: products"]
  R-354296883["name: test-topic"]
  end
  subgraph MySqlTable
  R-623545925["name: products"]
  R1664761355["name: test-topic"]
  end
  subgraph KafkaTopic
  R-41078963["clientOverrides: group.id: hoptima...
name: products
numPartitions: null"]
  R1728328189["clientOverrides: group.id: hoptima...
name: test-topic
numPartitions: null"]
  end
  subgraph SqlJob
    R-41078963 --> R663951423
    R1728328189 --> R663951423
  end
  subgraph BrooklinCDC
    R-623545925 --> R1206753277
    R1664761355 --> R-354296883
  end
  subgraph MySqlTable
  end
  subgraph KafkaTopic
    R1206753277 --> R-41078963
    R-354296883 --> R1728328189
  end

Loading

[vmaheshw]

In the mermaid drawing, at the datastream level, it uses the name of the MYSQL table. Does this mean the name of the data stream created and the MYSQL table is the same? Since, auto-generation is involved, it will probably make more sense to prefix the name with the object type.

Similarly, it shows overrides at the next step. What does it mean and how is it useful? Is it showing the properties of the object at that level?

Lastly, it shows SQL job, rather than Flink SQL job. It is important since the planner is specifically generating a FlinkSQL job.