examples: add a postgres replication example using CDC

[rockwotj]

Note that we can't dynamically do anything with the schema at the
moment, we don't emit schema changes and creating the migration
statements would require a complicated sql_raw output.

[mihaitodor]

LGTM, just one small nitpick. Feel free to

[ligfx]

Two thoughts:

• If @operation == "begin" || @operation == "commit" isn't handled explicitly (i.e. they don't care about restraints, as many analytical warehouses don't), what do those messages look like? Will they mess up the following SQL code or are they just empty messages?

• Relying on @operation to choose INSERT and UPDATE is in general not safe and will do weird things if messages come out of order or get replayed. Usually, people will either do INSERT ... ON CONFLICT or MERGE to perform an upsert (and that's what the Kafka Connect JDBC Sink connector does in upsert mode).

  Possible approach using MERGE (requires Postgres >= 15):

sql_raw:
  driver: postgres
  dsn: postgres://michaelmaltese@localhost:5432/michaelmaltese?sslmode=disable
  args_mapping: |
    root = [ this.id, this.name, this.updated_at ]
  query: |
    MERGE INTO journey_apps3_cdc AS old
    USING (SELECT
      $1::integer id,
      $2 name,
      $3::timestamptz updated_at,
    ) AS new
    ON new.id = old.id
    WHEN MATCHED THEN
      UPDATE SET
        name = case when new.updated_at > old.updated_at OR old.updated_at is null THEN new.name ELSE old.name END,
        updated_at = greatest(new.updated_at, old.updated_at)
    WHEN NOT MATCHED THEN
      INSERT (id, name, updated_at) VALUES (
        new.id,
        new.name,
        new.updated_at
      )
    ;

[ligfx]

Upsert comment above. Sorry, I'm a bit rusty with GitHub!

[rockwotj]

Will they mess up the following SQL code or are they just empty messages?

Yes they are empty and will mess up stuff downstream.

[rockwotj]

If you don't have foreign key restraints than batching by transaction is quite useless indeed.