Test EKS bulk import on Graviton

### Description / Background

Under parent issue:
- https://github.com/gchq/sleeper/issues/7065

We'd like an option to run bulk import with EKS on ARM architecture.

We assume we'll want this to be the default, although it would be good to compare.

### Technical Notes / Implementation Details

Spark creates pods for the executors with its Kubernetes integration:

https://spark.apache.org/docs/3.5.6/running-on-kubernetes.html

Fargate on EKS doesn't support Graviton.

EKS Auto Mode lets you configure the architecture for a node pool:
https://docs.aws.amazon.com/cdk/api/v2/docs/aws-cdk-lib.aws_eks_v2-readme.html#eks-auto-mode
https://docs.aws.amazon.com/eks/latest/userguide/create-node-pool.html

#### Testing

We can compare the performance with a large bulk import job, e.g. 1 billion rows, between ARM and x86.

### Dependencies / Blockers

This seems likely to conflict with the following issue, as both work on the same CDK code:
- https://github.com/gchq/sleeper/issues/7070

It's possible that the following issue may also involve similar CDK changes:
- https://github.com/gchq/sleeper/issues/7066

The following issue implemented this on EKS Auto Mode, but didn't specifically test performance with Graviton:
- https://github.com/gchq/sleeper/issues/7668

Start with the CDK code related to EKS and the linked Spark Kubernetes integration and EKS Auto Mode documentation. Configure an ARM node pool for bulk import, then run a large job such as 1 billion rows on both Graviton and x86 and compare the performance.

Written by the indexing model from the issue text.