Write backfill script for Airflow parent runs

[collado-mike]

As described in OpenLineage/OpenLineage#664 , the OpenLineage Airflow implementation has always sent back the ParentRunFacet with less than correct information in the runId and job name fields. It's possible to backfill this information in existing databases, which is necessary for a complete and correct implementation of #1928 .

The approach would be to split the parent job name on the . character to determine the DAG name and concatenate that with the runId (usually something like scheduled__2022-03-14T01:40:10+00:00) to generate UUID (see UUIDv3 generation in the OpenLineage implementation here). Note that some job names have the DAG followed by the task group, then the task id, all separated by . characters, so we should choose the left-most name of the task id.

[conorbev]

Done