[FEATURE] Flint index state management

[penghuo]

• CREATE INDEX, Spark streaming Job,

    1. CREATE NEW DOC in .query_execution_request_{mys3}, write docId into system.env.
    2. Flint-Extension, read system.env.

  type: streaming
    state = "running"
    jobId = “job1”
    appId = “app1”
    lastupdateTime = 0

• RECOVER INDEX

    1. CREATE NEW DOC in .query_execution_request_{mys3}, write docId into system.env.
    2. Flint-Extension, read system.env.

    docID = FlintExt.state().docId()
    if (state == "running") {
      // occ
      updateState(indices, docId): {
        type: streaming
        state = "running"
        jobId = “job2”
        appId = “app2”
        lastupdateTime = 0
      }
    }

• MANUAL REFRESH

    1. CREATE NEW DOC in .query_execution_request_{mys3}, write docId into system.env.
    2. Flint-Extension, read system.env.

    type: streaming
      state = "running"
      jobId = “job3”
      appId = “app3”
      lastupdateTime = 0

[penghuo]

• index: read fome spark.flint.job.requestIndex
• docId is stored in flint-index-metadata.
• streaming doc

type: streaming
state = "running"
jobId = “jobId”
appId = “appId”
lastupdateTime = 0

[dai-chen]

Closing. RECOVER command changes will be covered in #57.