feat: convert Benchmarks API to use FastAPI router (#4309)

# What does this PR do?

Convert the Benchmarks API from @webmethod decorators to FastAPI router
pattern, matching the Batches API structure.

One notable change is the update of stack.py to handle request models in
register_resources().

Closes: #4308 

## Test Plan

CI and `curl http://localhost:8321/v1/inspect/routes | jq '.data[] |
select(.route | contains("benchmark"))'`

---------

Signed-off-by: Sébastien Han <seb@redhat.com>

docs/static/llama-stack-spec.yaml

@@ -9166,20 +9166,23 @@ components:
           type: string
           const: benchmark
           title: Type
+          description: The resource type, always benchmark.
           default: benchmark
         dataset_id:
           type: string
           title: Dataset Id
+          description: Identifier of the dataset to use for the benchmark evaluation.
         scoring_functions:
           items:
             type: string
           type: array
           title: Scoring Functions
+          description: List of scoring function identifiers to apply during evaluation.
         metadata:
           additionalProperties: true
           type: object
           title: Metadata
-          description: Metadata for this evaluation task
+          description: Metadata for this evaluation task.
       type: object
       required:
       - identifier
@@ -9195,10 +9198,12 @@ components:
             $ref: '#/components/schemas/Benchmark'
           type: array
           title: Data
+          description: List of benchmark objects.
       type: object
       required:
       - data
       title: ListBenchmarksResponse
+      description: Response containing a list of benchmark objects.
     BenchmarkConfig:
       properties:
         eval_candidate:
@@ -9848,6 +9853,45 @@ components:
       - $ref: '#/components/schemas/RowsDataSource'
         title: RowsDataSource
       title: URIDataSource | RowsDataSource
+    RegisterBenchmarkRequest:
+      properties:
+        benchmark_id:
+          type: string
+          title: Benchmark Id
+          description: The ID of the benchmark to register.
+        dataset_id:
+          type: string
+          title: Dataset Id
+          description: The ID of the dataset to use for the benchmark.
+        scoring_functions:
+          items:
+            type: string
+          type: array
+          title: Scoring Functions
+          description: The scoring functions to use for the benchmark.
+        provider_benchmark_id:
+          anyOf:
+          - type: string
+          - type: 'null'
+          description: The ID of the provider benchmark to use for the benchmark.
+        provider_id:
+          anyOf:
+          - type: string
+          - type: 'null'
+          description: The ID of the provider to use for the benchmark.
+        metadata:
+          anyOf:
+          - additionalProperties: true
+            type: object
+          - type: 'null'
+          description: The metadata to use for the benchmark.
+      type: object
+      required:
+      - benchmark_id
+      - dataset_id
+      - scoring_functions
+      title: RegisterBenchmarkRequest
+      description: Request model for registering a benchmark.
     AllowedToolsFilter:
       properties:
         tool_names:
@@ -11053,6 +11097,33 @@ components:
       - batch_id
       title: CancelBatchRequest
       type: object
+    ListBenchmarksRequest:
+      description: Request model for listing benchmarks.
+      properties: {}
+      title: ListBenchmarksRequest
+      type: object
+    GetBenchmarkRequest:
+      description: Request model for getting a benchmark.
+      properties:
+        benchmark_id:
+          description: The ID of the benchmark to get.
+          title: Benchmark Id
+          type: string
+      required:
+      - benchmark_id
+      title: GetBenchmarkRequest
+      type: object
+    UnregisterBenchmarkRequest:
+      description: Request model for unregistering a benchmark.
+      properties:
+        benchmark_id:
+          description: The ID of the benchmark to unregister.
+          title: Benchmark Id
+          type: string
+      required:
+      - benchmark_id
+      title: UnregisterBenchmarkRequest
+      type: object
     DialogType:
       description: Parameter type for dialog data with semantic output labels.
       properties: