feat: Add guest execution agent and vsock support

- Introduced a standalone guest execution agent to facilitate command execution within the guest VM over Firecracker's vsock.
- Enhanced FirecrackerClient to support vsock configuration, validating guest CID and UDS path.
- Updated configuration to include vsock settings by default.
- Refactored FirecrackerProcessManager to manage runtime sockets more effectively.
- Implemented guest agent installation and provisioning in the image store.
- Added API endpoints for executing commands in the VM and installing the guest agent.
- Improved error handling for guest execution failures and agent availability.

uvm/firecracker/process.py

@@ -124,7 +124,7 @@ class FirecrackerProcessManager:
             ) from error
 
         socket_path = Path(vm.socket)
-        self._remove_socket(socket_path)
+        self._remove_runtime_sockets(vm)
 
         log_path = Path(vm.log)
         process: _PopenLike | None = None
@@ -275,14 +275,16 @@ class FirecrackerProcessManager:
         if vm.pid is None:
             if vm.spawn_pending or vm.operation == "spawning" or vm.process_start_time is not None:
                 return "unknown"
-            try:
-                Path(vm.socket).lstat()
-            except FileNotFoundError:
-                return "gone"
-            except OSError:
+            for socket_path in (Path(vm.socket), self._settings.guest_socket_path(vm.id)):
+                try:
+                    socket_path.lstat()
+                except FileNotFoundError:
+                    continue
+                except OSError:
+                    return "unknown"
+                # An unowned socket may belong to a child whose identity was never saved.
                 return "unknown"
-            # An unowned socket may belong to a child whose identity was never saved.
-            return "unknown"
+            return "gone"
 
         stat = self._process_stat(vm.pid)
         if stat is None:
@@ -330,14 +332,14 @@ class FirecrackerProcessManager:
                 f"refusing to stop Firecracker PID {vm.pid}: process identity is ambiguous"
             )
         if observation == "gone":
-            self._remove_socket(Path(vm.socket))
+            self._remove_runtime_sockets(vm)
             return
 
         try:
             pidfd = _open_pidfd(vm.pid)
         except ProcessLookupError:
             self._reap_child(vm.pid)
-            self._remove_socket(Path(vm.socket))
+            self._remove_runtime_sockets(vm)
             return
         except (OSError, NotImplementedError) as error:
             raise FirecrackerError(
@@ -369,7 +371,7 @@ class FirecrackerProcessManager:
                                 "process identity and socket retained"
                             )
             self._reap_child(vm.pid)
-            self._remove_socket(Path(vm.socket))
+            self._remove_runtime_sockets(vm)
         except (OSError, NotImplementedError) as error:
             raise FirecrackerError(
                 f"could not safely stop Firecracker process {vm.pid}: {error}"
@@ -455,6 +457,10 @@ class FirecrackerProcessManager:
         if process is not None and process.poll() is not None:
             self._children.pop(pid, None)
 
+    def _remove_runtime_sockets(self, vm: VmRecord) -> None:
+        self._remove_socket(self._settings.guest_socket_path(vm.id))
+        self._remove_socket(Path(vm.socket))
+
     @staticmethod
     def _remove_socket(socket_path: Path) -> None:
         try: