Define GPU-CR backend in API and regenerate protos (#135)

Add GpuCrBackendConfig to snapshot_agent.proto and BackendConfig oneof. Regenerate Go and Python proto bindings.

Author: Edwinhr716

pkg/client/python/pyproject.toml

@@ -9,13 +9,13 @@ description = "Timeslice library including Snapshot Agent client"
 readme = "README.md"
 requires-python = ">=3.8"
 dependencies = [
-    "grpcio>=1.82.1",
-    "protobuf>=5.26.1",
+    "grpcio>=1.83.0",
+    "protobuf>=7.35.1",
 ]
 
 [project.optional-dependencies]
 dev = [
-    "grpcio-tools>=1.82.1",
+    "grpcio-tools>=1.83.0",
 ]
 
 [tool.setuptools.packages.find]

pkg/client/python/timeslice/snapshot_agent/snapshot_agent_pb2.py

@@ -5,7 +5,7 @@
 
 from timeslice.snapshot_agent import snapshot_agent_pb2 as timeslice_dot_snapshot__agent_dot_snapshot__agent__pb2
 
-GRPC_GENERATED_VERSION = '1.82.1'
+GRPC_GENERATED_VERSION = '1.83.0'
 GRPC_VERSION = grpc.__version__
 _version_not_supported = False
 
@@ -101,6 +101,10 @@ def WorkloadChannel(self, request_iterator, context):
         (no HTTP server) use this instead of AppEndpointConfig. The first message
         from the workload must be a RegisterWorkload; subsequent messages are
         CommandResults acknowledging AgentCommands.
+
+        Registration is node-scoped: the workload registers with the agent on
+        its own node, and Snapshot/Restore requests carrying an AppChannelConfig
+        must be sent to that same agent.
         """
         context.set_code(grpc.StatusCode.UNIMPLEMENTED)
         context.set_details('Method not implemented!')