Closes #819

How it was tested locally:

application.properties changed properties:

notification.enabled=true
notification.sender.address=<redacted>
spring.mail.host=<redacted>
spring.mail.port=<redacted>
spring.mail.properties[mail.smtp.connectiontimeout]=5000
spring.mail.properties[mail.smtp.timeout]=5000
spring.mail.properties[mail.smtp.writetimeout]=5000

sparkYarnSink.hadoopResourceManagerUrlBase=http://<redacted.:8088

db.driver=org.postgresql.Driver
db.url=jdbc:postgresql://localhost:5432/hyperdriver
db.user=hyperdriver
db.password=hyperdriver

• Create dummy workflow, click run
• Job Instance should be stuck in Submitting status.
• Create a suitable notification rule (E.g. receive for all projects and prefixes). Don’t forget to activate it
• Copy a real executor job id from the test system and copy it to your job instance. Also set the job status to ‘Running’. E.g.
  UPDATE job_instance SET (job_status, executor_job_id) VALUES (‘Running’,<real-executor-job-id>) WHERE id = <your-local-job-instance-id>
• Refresh job instance table (directly in DB), see stack trace filled in the diagnostics column of your job instance.
• Wait a couple of minutes, receive e-mail notification with stacktrace.

I did it both for failed and successful runs, and also for a failed run whose application page was already empty

After these tests were finished, finally I executed

CALL archive_dag_instances(
        i_to_ts => now()::timestamp,
        i_max_records => 200000,
        i_chunk_size => 10000
     );