Reorganize OpenMetadata Python SDK

[TeddyCr]

Context Python SDK for OpenMetadata contains many different operations for interacting with the OpenMetadata API.

Generally speaking, operations are separated based on resources, and the entire Ometa class uses inheritance.

While this proves convenient, it also creates noise for user as they can see a mix of operations when instantiating the Ometa object.

Suggested Approach To streamline this approach we suggest different improvements:

• move from inheritance to composition (each object receiving an instance of the client to interact with the server)
• structure the SDK by resources, client, error, and utils with a core resources grouping generic operations such as get_by_name, list, post, etc.
• auto generate SDK documentation for clear communication
• annotate private vs public methods

[pmbrull]

brain dump

# "PRIVATE"
class _OmetaTestClient:

    def __init__(config, Core)
          config = config

    def upload_test()
         ...

    def ...
       create_or_update/get/list ->> This is from core

# "PUBLIC"
class OmetaCore:

     def __init__(config)
          config = config

     ## DOCS
     def get_...
     def create_or_update
     def list...
     ## !DOCS

class OmetaAPI:
     core = OmetaCore()
     _test = OmetaTestClient(core)

     def list = core.list() ->> define alias

api = OmetaAPI()

api.core.list?
api.list ->>

api._test.upload
api._lineage.get

# "Interal clients"
ometa.__test.upload_tests()

# "External clients"
TestClient()

class OmetaTestClient(OmetaCore)

tables = Table(client)
Listed = tables.list()

Ometa -> list, get put
TestClient -> upload test
TableClient -> upload Sample Data?