Improve performance of the structured data generator

[hariso]

Feature description

The following pipeline can produce 6-7k msg/s (on my machine and on a c7i.xlarge EC2 instance). That is pretty slow when compared to real data sources, such as Postgres (which can achieve 10x more).

One possibility is always to return one pre-generated record. That's what we also do with the file generator.

Example pipeline:

version: "2.2"
pipelines:
  - id: generator-kafka
    status: running
    connectors:
      - id: generator
        type: source
        plugin: builtin:generator
        settings:
          collections.users.format.type: structured
          collections.users.format.options.id: int
          collections.users.format.options.name: string
          collections.users.format.options.email: string
          collections.users.format.options.position: string
          collections.users.format.options.salary: int
          collections.users.format.options.full_time: bool
          collections.users.format.options.hire_date: time
          collections.users.format.options.created_at: time
          collections.users.format.options.updated_at: time
          sdk.schema.extract.key.enabled: false
          sdk.schema.extract.payload.enabled: false
      - id: kafka-destination
        type: destination
        plugin: builtin:kafka
        name: kafka-destination
        settings:
          servers: "benchi-kafka:9092"
          topic: "generator.to.kafka"
          compression: "none"