feat: Add InferredSchemaLoader for runtime schema inference

Add a new InferredSchemaLoader component that infers JSON schemas by reading
a sample of records from the stream at discover time. This enables declarative
connectors to automatically generate schemas for streams where the schema is
not known in advance or changes dynamically.

Key features:
- Reads up to record_sample_size records (default 100) from the stream
- Uses SchemaInferrer to generate JSON schema from sample records
- Handles errors gracefully by returning empty schema
- Fully integrated with declarative component schema and model factory
- Includes comprehensive unit tests

Requested by: AJ Steers ([email protected]) @aaronsteers

Co-Authored-By: AJ Steers <[email protected]>

Author: devin-ai-integration[bot]

airbyte_cdk/sources/declarative/declarative_component_schema.yaml

@@ -1548,13 +1548,15 @@ definitions:
           loaders defined first taking precedence in the event of a conflict.
         anyOf:
           - "$ref": "#/definitions/InlineSchemaLoader"
+          - "$ref": "#/definitions/InferredSchemaLoader"
           - "$ref": "#/definitions/DynamicSchemaLoader"
           - "$ref": "#/definitions/JsonFileSchemaLoader"
           - title: Multiple Schema Loaders
             type: array
             items:
               anyOf:
                 - "$ref": "#/definitions/InlineSchemaLoader"
+                - "$ref": "#/definitions/InferredSchemaLoader"
                 - "$ref": "#/definitions/DynamicSchemaLoader"
                 - "$ref": "#/definitions/JsonFileSchemaLoader"
                 - "$ref": "#/definitions/CustomSchemaLoader"
@@ -2462,6 +2464,40 @@ definitions:
       $parameters:
         type: object
         additionalProperties: true
+  InferredSchemaLoader:
+    title: Inferred Schema Loader
+    description: Infers a JSON Schema by reading a sample of records from the stream at discover time. This is useful for streams where the schema is not known in advance or changes dynamically.
+    type: object
+    required:
+      - type
+      - retriever
+    properties:
+      type:
+        type: string
+        enum: [InferredSchemaLoader]
+      retriever:
+        title: Retriever
+        description: Component used to coordinate how records are extracted across stream slices and request pages.
+        anyOf:
+          - "$ref": "#/definitions/SimpleRetriever"
+          - "$ref": "#/definitions/AsyncRetriever"
+          - "$ref": "#/definitions/CustomRetriever"
+      record_sample_size:
+        title: Record Sample Size
+        description: The maximum number of records to read for schema inference. Defaults to 100.
+        type: integer
+        default: 100
+        example:
+          - 100
+          - 500
+          - 1000
+      stream_name:
+        title: Stream Name
+        description: The name of the stream for which to infer the schema. If not provided, it will be inferred from the stream context.
+        type: string
+      $parameters:
+        type: object
+        additionalProperties: true
   InlineSchemaLoader:
     title: Inline Schema Loader
     description: Loads a schema that is defined directly in the manifest file.