Add / use poll.interval.ms config option (!= scylla.query.time.window.size)

[hartmut-co-uk]

Description

To allow to tune/customise the behaviour of one's source connector setup, I'd like to also have a config option poll.interval.ms in addition to scylla.query.time.window.size which defines effectively the query time window size + query interval for a 'live' / caught up worker task.

As per my understanding / reasoning the poll.interval.ms would/should be smaller than scylla.query.time.window.size - with the latter being applied while catching up / init phase.

Workers (connect tasks) ideally will evenly scatter queries for to the assigned array of streamIds / streamIdGroups (scylla-cdc-java worker task?).

Config Field Definition

poll.interval.ms
Positive integer value that specifies the frequency in milliseconds the connector should wait to poll for new data in each worker task (Vnode). Defaults to 15.000 milliseconds.

• Type: Integer
  Importance: Low
  Default: 15000
  Frequency in ms to poll for new data in each table.

References

• Improves/Fixes TableBackedProgressManager scylla-cdc-go#10 (comment)