Add missing instance types and fix resource specs

[FarhanTejani]

What's changing and why?

Adding missing instance types to HyperpodInstanceType enum and INSTANCE_RESOURCES, and fixing incorrect resource specs for existing types.

HyperpodInstanceType enum — added 32 missing types:

• c6i family (large through 32xlarge)
• m6i family (large through 32xlarge)
• r6i family (large through 32xlarge)
• ml.p5.4xlarge, ml.p6-b200.48xlarge, ml.p6-b300.48xlarge, ml.p6e-gb200.36xlarge
• ml.trn2.3xlarge

INSTANCE_RESOURCES — new entries and fixes:

• Added ml.p6-b300.48xlarge (cpu=192, gpu=8, memory=4096, efa=16)
• Added ml.trn2.3xlarge (cpu=12, trainium=1, memory=128, efa=1)
• Fixed ml.p6-b200.48xlarge memory: 2024 → 2048
• Fixed ml.trn2.48xlarge EFA count: 0 → 16

All specs verified via aws ec2 describe-instance-types.

Helm charts — added ml.p6-b300.48xlarge to:

• nvidia-device-plugin node affinity (values.yaml)
• aws-efa-k8s-device-plugin supported instances (values.yaml)
• health-monitoring-agent node affinity

Before/After UX

Before: Users cannot create training jobs on c6i, m6i, r6i, p5.4xlarge, p6-b200, p6-b300, p6e-gb200, or trn2.3xlarge instances. EFA configuration is blocked for trn2.48xlarge despite hardware support.

After: All listed instance types are supported. EFA resource allocation works correctly for trn2 instances.

How was this change tested?

• All unit tests pass (1026 passed)
• Instance specs verified via aws ec2 describe-instance-types across multiple regions

Are unit tests added?

No new tests required — the enum and resource map are covered by existing validator tests.

Are integration tests added?

N/A

Reviewer Guidelines

‼️ Merge Requirements: PRs with failing integration tests cannot be merged without justification.

One of the following must be true:

• All automated PR checks pass
• Failed tests include local run results/screenshots proving they work
• Changes are documentation-only