fix(ec2): include supported usage classes in DescribeInstanceTypes

[luiseterc]

Summary

Fixes the EC2 DescribeInstanceTypes response so it includes the supported usage classes required by Karpenter and other AWS-compatible clients. Fixes #3346.

Type of change

• Bug fix (fix:)
• New feature (feat:)
• Breaking change (feat!: or fix!:)
• Docs / chore

AWS Compatibility

Floci previously omitted supportedUsageClasses from DescribeInstanceTypes. Karpenter 1.8.8 interprets the missing field as no compatible capacity types and rejects on-demand NodeClaims before CreateFleet; the same direct CreateFleet request succeeds. The catalog now returns on-demand and spot, and the query handler serializes both values using the AWS EC2 Query response shape.

The change covers m5.large, t4g.medium, and m6gd.large, the instance types exercised by the compatibility gate.

Changes

• Add validated supportedUsageClasses data to the instance-type catalog, defaulting to on-demand and spot.
• Serialize the list as <supportedUsageClasses><item>...</item></supportedUsageClasses> in DescribeInstanceTypes.
• Add an SDK integration test asserting the values for representative amd64 and arm64 instance types.

Validation

• ./mvnw -q -DskipTests package
• ./mvnw -q -Dtest=Ec2IntegrationTest test

Checklist

• ./mvnw test passes locally (targeted EC2 integration test and package build pass; full suite not run)
• New or updated integration test added
• Commit messages follow Conventional Commits

[greptile-apps]

Greptile Summary

Adds AWS-compatible supported usage classes to EC2 DescribeInstanceTypes.

• Defaults catalog entries to on-demand and spot and validates the field.
• Projects the values into instance-type response maps and serializes the EC2 Query XML list.
• Adds AWS SDK compatibility coverage for representative amd64 and arm64 instance types.
• Includes an unrelated Step Functions test-comment clarification with no behavioral effect.
• The required greptile_confidence label could not be added because GitHub CLI authentication is unavailable.

Confidence Score: 5/5

The PR appears safe to merge with no outstanding correctness, compatibility, or repository-rule issues.

Catalog defaults survive YAML deserialization, every described instance type receives the projected values, the XML structure matches the AWS Query list shape, and SDK compatibility coverage verifies the response.

Important Files Changed

Filename	Overview
src/main/java/io/github/hectorvent/floci/services/ec2/Ec2InstanceTypeCatalog.java	Adds validated default usage-class metadata and includes it in instance-type response projections.
src/main/java/io/github/hectorvent/floci/services/ec2/Ec2QueryHandler.java	Emits supported usage classes using the AWS EC2 Query list shape.
compatibility-tests/sdk-test-java/src/test/java/com/floci/test/Ec2Tests.java	Verifies AWS SDK decoding for three representative instance types.
src/test/java/io/github/hectorvent/floci/services/stepfunctions/AslExecutorFailStateErrorCauseTest.java	Clarifies fixture documentation without changing test behavior.

_{Reviews (8): Last reviewed commit: "test(ec2): document usage class regressi..." | Re-trigger Greptile}

[hectorvent]

@luiseterc main is broken at the moment. PR #3345 will fix it, you can rebase main as soon this PR get merged.

[luiseterc]

@hectorvent The current Build and Test failure is isolated to shard 3 and is unrelated to this EC2 PR. CloudTrailLogWriterBoundedRetryIntegrationTest.retryStateStaysBoundedAndDeliversRetainedPrefixAfterDestinationRecovery fails at line 85 because it expects events/0000.txt but receives events/1000.txt. The test retains 1,024 records across two S3 log objects, while the object names include random suffixes; deliveredRecords() concatenates records in S3 listing order and therefore assumes an ordering the keys do not guarantee. Shards 1, 2, and 4, native builds, compatibility tests, and repository checks all pass. This looks like a CloudTrail test-ordering flake from the existing implementation in #3298, not a regression from #3347.