fix(dbt): add use_identifiers option and avoid duplicate descriptions

[remisalmon]

Checklist

• The PR conforms to DataHub's Contributing Guideline (particularly Commit Message Format)
• Links to related issues (if applicable)
• Tests for the changes have been added/updated (if applicable)
• Docs related to the changes have been added/updated (if applicable)

This PR adds an additional option use_identifiers to the dbt configuration to use nodes identifiers instead of names if those are defined, and default back to names if not.

Previously this was achieved by setting load_schemas=False however there are cases when we want to load schemas while using node identifiers instead of names (for ex. dbt does not write column descriptions to Snowflake on views* and those need to be ingested with load_schemas=True, while using identifiers to preserve the lineage).

*see https://github.com/dbt-labs/dbt/issues/3291

Also:

• renamed load_catalog to load_schemas in dbt.py for consistency with the config file
• fixed a case issue when reading columns from catalog vs manifest files (names are upper case in the catalog but lower case in the manifest...)
• added a check for node/column description and comment to be different if both are displayed

[shirshanka]

LGTM!