Add the most obvious Exception types and good logging

[blythed]

To do initially list of important Exceptions to handle. Particular focus on integrations.

• Add exceptions into external units like server requests, db connections, metadata connection, file writings, etc #1336
• Add exceptions into AI related errors, model fit, predict problems, etc #1337
• Add exceptions to components, ModelException, EncoderException, VectorIndexException, etc #1338
• Confirm config hash handshake when client sends request to service (vector-search, cdc) #1322

[fnikolai]

• When interacting with any external system (mongo, ibis, dask), add an info log, and then a success log. Your may use: debug(), info(), warn(), error()

from superduperdb import logging.

logging.info("Connecting to: <type of service}")
.... connect() ...
logging.success("Connected to: ", <client_info>)

• When submitting a request to an external system (mongo, ibis, dask), add an info log, and then a success log.

from superduperdb import logging.

logging.info("Submit job: <job description}")
.... submit() ...
logging.success("Submitted job: ", <job_id or future>)

• Wrap interactions with external systems (mongo, ibis, dask), into a try-catch. Fail-fast if an exception happens.

  try:
      databackend = build(cfg.data_backend, data_backends)
      logging.info("Data Client is ready.", databackend.conn)
  except Exception as e:
      # Exit quickly if a connection fails.
      logging.error("Error initializing to DataBackend Client:", str(e))
      sys.exit(1)