[Bug]: Granules do not get persisted to the catalog under heavy load

[ngachung]

Under heavy load, lambdas writing to RDS/Elasticsearch fail due to KnexTimeoutError and ConnectionTimeout

[ngachung]

On 3/3/25, we made the following changes to production MDPS DS deployment that should be better at handling operational load.

1. Increase lambda max concurrent execution from 1,000 to 3,000.
2. Update Aurora PostgreSQL serverless maximum capacity from 4 ACUs to 16 ACUs (which allows for 3,360 maximum connections)
3. Change DS Elasticsearch from 1 node c6g.large.elasticsearch to 3 node c6g.xlarge.elasticsearch