Improve pg_replication docs

[VioletM]

1. The example pipeline in pg_replication source is confusing. The example pipeline in pg_replication_pipeline.py is confusing. It includes creating tables and is good for local testing, but hard to use with the real world scenario. We need to clean it up and add the typical example of initial load and then replication
2. To add tables to publication it requires the owner rights of the table
3. Describe in more details what is happening in the init_replication and how to set up the replication to avoid data loss