In HELP-57694, a customer attempted to kill a chunk migration by interrupting the _recvChunkStart command. This didn't actually kill the migration, but it did bring to light that the migration was stuck due to the _recvChunkStart command waiting for a write concern that it couldn't fulfil (in this case, the write concern was set to w: 3 and only 2 nodes were up).

The purpose of this ticket is to investigate whether there should be some timeout attached to the _recvChunkStart command to prevent this.