[Data] Support to read from/ write to hologres in ray

Description

Hologres is a real-time interactive analytics engine that is compatible with PostgreSQL, developed by Alibaba Cloud(website). The default read/write interface cannot distribute the operation in parallel for hologres, so new APIs with optimizations specific to hologres can be helpful for hologres users, especially working on large scale datasets.

Use case

With new APIs, users can read from/ write to hologres in an efficient way in ray, typical use cases are like

ds = ray.data.read_hologres(
     schema_name="public",
     table_name="my_table",
     connection_uri="postgresql://username:password@host:port/database",
     ...
)

ds.write_hologres(
      schema_name="public",
      table_name="my_table",
      connection_uri=hologres_connection_uri,
      ...
)

Start by reviewing Ray Data's existing read and write APIs and the proposed read_hologres and write_hologres entry points. Clarify the Hologres-specific parallelization and connection requirements, then define the API behavior and tests needed to show efficient reads and writes for the example use cases.