feat: add InstancePools to EndpointConfig ProductionVariants

Bumps the sagemaker SDK to v1.244.0 to expose the InstancePools field on
ProductionVariant, enabling heterogeneous endpoints that provision across a
prioritized list of instance types. The field maps automatically as a nested
field on the already-generated ProductionVariant shape; no hooks needed.

The SDK bump also introduces unrelated new resources and fields. To keep the
change scoped to InstancePools, new resources (AIBenchmarkJob,
AIRecommendationJob, AIWorkloadConfig, MlflowApp, PresignedMlflowAppUrl) are
added to ignore.resource_names, and new out-of-scope fields (MetricsConfig,
Specifications, ScaleInPolicy, DataCacheConfig, SchedulingConfig,
ModelPackageRegistrationType, AdditionalModelDataSources, BaseModel,
DatasetSource, MlflowConfig, ModelPackageConfig, ServerlessJobConfig,
VariantInstanceProvisionTimeoutInSeconds, MlflowExperimentName, and others)
are added to ignore.field_paths on both their input and output shapes.

Author: michaelhtm

ATTRIBUTION.md

@@ -1562,9 +1562,7 @@ Subdependencies:
 * `github.com/aws/aws-sdk-go-v2/internal/endpoints/v2`
 * `github.com/aws/smithy-go`
 
-### github.com/aws/smithy-go
 
-License Identifier: Apache-2.0
 
 ### github.com/ghodss/yaml
 

apis/v1alpha1/ack-generate-metadata.yaml

@@ -1,13 +1,13 @@
 ack_generate_info:
-  build_date: "2026-07-08T00:53:44Z"
-  build_hash: 0a6b791fa10a6140d862be334a6891868ee588eb
-  go_version: go1.26.0
-  version: v0.61.0
-api_directory_checksum: c4e5b363f35aec7f0c52278a9b963e41bb4a5fd5
+  build_date: "2026-07-16T18:47:41Z"
+  build_hash: 01fb153549af6a2df3747478105143e989ff4355
+  go_version: go1.26.4
+  version: v0.61.0-1-g01fb153
+api_directory_checksum: 0e028d4d41dad2e2a78c39a23c692776ccc01f8b
 api_version: v1alpha1
-aws_sdk_go_version: v1.39.2
+aws_service_sdk_version: v1.244.0
 generator_config_info:
-  file_checksum: 6a72988a72fb9638b541c8bec98d4d811d29b64d
+  file_checksum: 2539bc87ce7580141047eea388b9d84de9f40f26
   original_file_name: generator.yaml
 last_modification:
   reason: API generation

apis/v1alpha1/enums.go

@@ -1275,6 +1275,9 @@ resources:
 ignore:
   resource_names:
     - Algorithm
+    - AIBenchmarkJob
+    - AIRecommendationJob
+    - AIWorkloadConfig
     # - App
     - AutoMLJob
     - AutoMLJobV2
@@ -1309,6 +1312,7 @@ ignore:
     - InferenceExperiment
     # - LabelingJob
     # - Model
+    - MlflowApp
     - MlflowTrackingServer
     - ModelCard
     - ModelCardExportJob
@@ -1326,6 +1330,7 @@ ignore:
     # - Pipeline
     - PresignedDomainUrl
     - PresignedNotebookInstanceUrl
+    - PresignedMlflowAppUrl
     - PresignedMlflowTrackingServerUrl
     # - ProcessingJob
     # - Project
@@ -1399,3 +1404,34 @@ ignore:
     - AdditionalS3DataSource.ETag
     - ModelPackageContainerDefinition.ModelDataETag
     - SourceAlgorithm.ModelDataETag
+    # Fields introduced by the v1.244.0 SDK bump that are out of scope for the
+    # InstancePools change. Ignored on both the Create input and Describe output
+    # shapes so they are excluded from Spec and Status rather than being moved to
+    # Status (which would happen if only the input path were ignored). Shape-level
+    # paths (e.g. ProductionVariant.X) cover the field wherever the shape is used.
+    - CreateEndpointConfigInput.MetricsConfig
+    - DescribeEndpointConfigOutput.MetricsConfig
+    - ProductionVariant.VariantInstanceProvisionTimeoutInSeconds
+    - ProductionVariantManagedInstanceScaling.ScaleInPolicy
+    - CreateInferenceComponentInput.Specifications
+    - DescribeInferenceComponentOutput.Specifications
+    - InferenceComponentSpecification.DataCacheConfig
+    - InferenceComponentSpecification.InstanceType
+    - InferenceComponentSpecification.SchedulingConfig
+    - InferenceComponentSpecificationSummary.DataCacheConfig
+    - InferenceComponentSpecificationSummary.InstanceType
+    - InferenceComponentSpecificationSummary.SchedulingConfig
+    - CreateModelPackageInput.ModelPackageRegistrationType
+    - DescribeModelPackageOutput.ModelPackageRegistrationType
+    - ModelPackageContainerDefinition.AdditionalModelDataSources
+    - ModelPackageContainerDefinition.BaseModel
+    - ModelPackageContainerDefinition.IsCheckpoint
+    - DataSource.DatasetSource
+    - StartPipelineExecutionInput.MlflowExperimentName
+    - CreateTrainingJobInput.MlflowConfig
+    - DescribeTrainingJobOutput.MlflowConfig
+    - CreateTrainingJobInput.ModelPackageConfig
+    - DescribeTrainingJobOutput.ModelPackageConfig
+    - TrainingJob.OutputModelPackageArn
+    - CreateTrainingJobInput.ServerlessJobConfig
+    - DescribeTrainingJobOutput.ServerlessJobConfig