Configuration setting so all connection errors are retried

[gregingenii]

Describe the feature

We're using dbt-spark to run tests on Databricks, and one of our tables has over 150 columns with tests associated. Each time we test at least a few will fail, randomly, due to a connection issue which looks to be transient.
The connector will only retry a connection if it considers the error 'retryable', usually that the cluster is pending start-up, so I would like to add a configuration entry to be able to override this behaviour and retry every connection issue. This would be a boolean we can add to the definition in profiles.yml called 'retry_all'

Who will this benefit?

Anyone that is hitting this random failure issue. DBT is unusable if we can't rely on the testing.

Are you interested in contributing this feature?

I have a fix that I'm using already, so I can open a PR to add this in. I would like to see if everyone thinks this is a good way to resolve this issue, and can adjust accordingly.