[BUG] RFS encounters OutOfMemoryError when processing an index with docs ~20MB in size

[AndreKurait]

What is the bug?

Customer used RFS to migrate a cluster with larger docs with default settings in Migration Assistant, encountering error

10:48:47,814 ERROR r.c.s.Schedulers [sharedSegmentReader-27] Scheduler 
worker in group main failed with an uncaught exception
java.lang.OutOfMemoryError: Java heap space

What are your migration environments?

ES 7.17 -> OS 1.3

How can one reproduce the bug?

Use RFS with snapshot with 20MB docs

What is the expected behavior?

RFS operates without OOM

Do you have any additional context?

[AndreKurait]

Note: There may be a workaround here to use "reindexFromSnapshotWorkerSize": "maximum" in the cdk.context.json to deploy RFS ECS instances vertically scaled (16 vCPU 32GB Memory).