drop subtasks in cancel-import/runner.c test (#1345)

This was causing a panic in Wasmtime when disposing of the guest task parent of
those subtasks.  Wasmtime shouldn't be panicking no matter what the guest does,
so I'll need to fix that, but meanwhile we might as well have good handle hygene
in our tests.

Signed-off-by: Joel Dice <joel.dice@fermyon.com>

Author: dicej

tests/runtime-async/async/cancel-import/runner.c

@@ -15,6 +15,7 @@ int main() {
     status = runner_subtask_cancel(subtask);
     assert(RUNNER_SUBTASK_STATE(status) == RUNNER_SUBTASK_RETURNED_CANCELLED);
     assert(RUNNER_SUBTASK_HANDLE(status) == 0);
+    runner_subtask_drop(subtask);
 
     runner_waitable_status_t status2 = test_future_void_write(writer);
     assert(RUNNER_WAITABLE_STATE(status2) == RUNNER_WAITABLE_DROPPED);
@@ -47,9 +48,12 @@ int main() {
     status = runner_subtask_cancel(subtask1);
     assert(RUNNER_SUBTASK_STATE(status) == RUNNER_SUBTASK_RETURNED_CANCELLED);
     assert(RUNNER_SUBTASK_HANDLE(status) == 0);
+    runner_subtask_drop(subtask1);
+
     status = runner_subtask_cancel(subtask2);
     assert(RUNNER_SUBTASK_STATE(status) == RUNNER_SUBTASK_STARTED_CANCELLED);
     assert(RUNNER_SUBTASK_HANDLE(status) == 0);
+    runner_subtask_drop(subtask2);
 
     // We still own the readable end of `reader2` and `writer2` since the
     // subtask didn't actually start, so drop it here.