Fix OpenLineage extraction for deferrable AthenaOperator

[kacpermuda]

Ol extraction in deferrable mode is not using output location from API response. This PR adjusts it, so that the behaviour is the same regardless of deferrable mode being on or off.
This PR also adds ExternalQueryRunFacet for Athena job and fixes the problem where the whole extraction failed if one of the columns in Athena was missing a description.

There is also something i'd like to discuss. Right now, the output datasets are different between START and COMPLETE events. When passing output_location="s3://<bucket>/dir/ to AthenaOperator,

in START event we get one output: Dataset(namespace="s3://<bucket>", name="dir")

in COMPLETE event we get two outputs: Dataset(namespace="awsathena://athena.eu-central-1.amazonaws.com", name="AwsDataCatalog.<db>.<table>") and Dataset(namespace="s3://<bucket>", name="dir/tables/2ee1bfeb-4c67-4f50-a49d-df5deeb5f034").

It's because in COMPLETE we are using the output from the API, so we know the exact location in S3 where the output has been saved so we replace whatever user has provided with what we get from the API. I wonder if it's something we expected and should stay like this or maybe we should somehow change that, to make it consistent in both events? I believe some people are using this first dataset based on this issue.

---

^ Add meaningful description above
Read the Pull Request Guidelines for more information.
In case of fundamental code changes, an Airflow Improvement Proposal (AIP) is needed.
In case of a new dependency, check compliance with the ASF 3rd Party License Policy.
In case of backwards incompatible changes please leave a note in a newsfragment file, named {pr_number}.significant.rst or {issue_number}.significant.rst, in newsfragments.

[kacpermuda]

We are still discussing the mismatch between output datasets, let's hold with merging this one.

[kacpermuda]

I've changed the Ol method to _on_complete() as most of the information needs the call to Athena anyway. I also removed the actual S3 location (that was sent in complete event) and replaced it with user provided prefix (that was sent in start event). I think it's ready now.