Tabular: Fix edge-case text preprocessing bug

[Innixma]

Issue #, if available:
resolves #2688

Description of changes:

• Previously the change added in [Tabular] Avoid copying identical dataframes #2532 created a subtle bug when only text features were present in the data due to the inplace renaming operation for raw text.
• Because we didn't test for this specific scenario where only text features are provided, we did not catch this issue.
• This fixes the bug by setting inplace=False during feature renaming
• Also adds a unit test to check that the bug is fixed.
• v0.6.2 workaround: Avoid only passing text features to TabularPredictor, instead include the text features with other features or create dummy features to pass alongside the text features.

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[github-actions]

Job PR-2705-d150e3e is done.
Docs are uploaded to http://autogluon-staging.s3-website-us-west-2.amazonaws.com/PR-2705/d150e3e/index.html