fix(tableau): make Tableau ingestor resilient to timeout exceptions

[skrydal]

While running Tableau ingestor against Tableau instances containing thousands of charts/dashboards which change often (hence also stale entity removal handler issues many soft-deletes) there is a risk of idle time between the last call to Tableau REST API to get metadata and sign-out call done at the time the source is closed. That time mostly is taken by transfomers and stale entity removal handler. In my experience when running the ingestor the time can grow to be even more than 10 minutes. If the ingestor operates within stable network there shouldn't be any particular problem but it happens that in my use-cases many times connection (requests.Session http pool) is dropped during the idle time. Note that problem here arises from no activity between Tableau client and server and connection drop is not initiated from either side but rather by networking layer in between, so there is no easy way to keep connection active without explicitly issuing requests. Since the sign-out is anyway only an attempt to finish gracefully, I simply added no-op exception handling during the closing.

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). If a new feature has been added a Usage Guide has been added for the same.
• For any breaking change/potential downtime/deprecation/big changes an entry has been made in Updating DataHub

[treff7es]

lgtm, thanks for the contribution