Optimize ReadN in CDC by collecting records in batches

[hariso]

Description

Closes #278.

Quick checks:

• I have followed the Code Guidelines.
• There is no other pull request for the same update/change.
• I have written unit tests.
• I have made sure that the PR is of reasonable size and can be easily reviewed.

[raulb]

@hariso Left some comments, but approving because I don't consider them blockers. I'm particularly interested in the benchmark results comparing to our latest tests due to:

1. Implementation is supposedly more performant actually collecting records in batches.
2. See if the change has any impact on #282

[hariso]

@hariso Left some comments, but approving because I don't consider them blockers. I'm particularly interested in the benchmark results comparing to our latest tests due to:

1. Implementation is supposedly more performant actually collecting records in batches.

2. See if the change has any impact on [Investigate: Why allocating memory is less efficient on `ReadN` #282](https://github.com/ConduitIO/conduit-connector-postgres/issues/282)

@raulb I've just run in on my EC2 instance (c7i.xlarge), and the CDC test that we currently have in streaming-benchmarks gives me cca. 64k msg/s, whereas this version of the connector gives around cca. 74k msg/s. I ran both versions twice.