Deployed NPDA loses connection to the database after a few days

[mbarton]

We need to port across this fix from E12: rcpch/rcpch-audit-engine#1004

It's not obvious how to to run this in Azure Container Apps. There are scheduled jobs which could write the token to the same Azure file share we store our config. But you configure them totally separately from the apps so the update job would have it's own managed identity and the main app would end up using a token from a different managed identity. They could perhaps share a user managed identity?

We could alternatively update the token on a cron within the docker container or a thread within the app.

[mbarton]

I think I’m leaning towards a thread to solve this as then it doesn’t matter how we deploy the container