Deploy multiple

[FrancoisPoinsot]

When there is too many connector to deploy, the distributed kafka-connect workers may have problem to handle the load.

Currently the cli only handle them one by one, so we resorted to to it in parallel in the bash itself.
But error handling and throttling it in bash is painful.
So why not do that in the cli.

Changes:

• In lib, add feature DeployMultipleConnector with a max parallel deployment fixed to 3.
  Multiple error will be aggregated and returned once all connectors configs given have been tried.

• Rework a bit the add the first unit tests. They are still not yet automatically executed.
  Mostly isolated the low level client which only implement the kafka-connect API from the higher level client which expose all the custom feature we use. Such as DeployMultipleConnector.

• In cli make the --file argument support referencing a folder.
  cli will try to parse every files in this folder. No subfolder will be searched.
  Goal is to reference a folder full of config to deploy and forget the parallelism/error handling on the bash side.

[FrancoisPoinsot]

every comment has been either answered or applied
please review

[fsilberstein]

All good! Great job 👍
When you create the release, maybe use a major version because of the breaking change for the flag name