Run the remote subprocess node in the execution mode a stock site uses

Test coverage only.

RemoteSubprocessTest pins execution_mode to queued in setUp(), and 32 of its 33 tests ran under it. The one that set synchronous asserted the cron sweep, never the node. So the feature's happy path was synchronous only in the sense that the stubbed remote had drained the child before answering - the local instance was queued throughout and every assertion came after a drainQueue().

The shipped default is synchronous, so on a stock site the node's whole lane runs inside drainInline(), in the request that started the run. Four tests now run its four outcomes there and drain no queue, because reaching the end is the assertion.

All four pass against the node as it stands, so this closes a coverage gap rather than a defect. Neutering the mode helper fails all four, which is what says they exercise the lane instead of repeating the queued ones.