Raise a check's store without durability and remove what earlier holders left (hq issue 306)

On the control node the store-bound packages of the controller's suite ran
five to seven times slower than on any other holder, and every controller
check that landed there ran past the suite's thirty minutes: a throwaway
store flushing to a disk the mesh's own store, bus and forge keep busy.
A store that lives for one check needs no crash safety.

A holder recreated mid-check left the check's store and bus running, and
the redelivery went to another machine, so nothing removed them: eleven
pairs across four machines. A starting holder has taken nothing, so every
container labelled with an ask of the seat is an earlier holder's.

cmd/mesh-builder/holder.go

@@ -274,6 +274,22 @@ func (h *holder) removeContainers(id string) (int, error) {
 	return builder.RemoveContainersOf(cleanup, h.remove, id)
 }
 
+// removeLeftBehind removes, once, what earlier holders on this machine left — called before anything is
+// taken, so none of it is this holder's — and says what it did. A runtime that cannot be asked is said
+// and the holder takes work all the same: what is left is waste, not a reason to stop building.
+func (h *holder) removeLeftBehind() int {
+	cleanup, stop := context.WithTimeout(context.Background(), killRemoves)
+	defer stop()
+	n, err := builder.RemoveLeftBehind(cleanup, h.remove)
+	switch {
+	case err != nil:
+		fmt.Fprintf(os.Stderr, "could not look for containers earlier builds left on %s: %v\n", h.on, err)
+	case n > 0:
+		fmt.Fprintf(os.Stderr, "removed %d container(s) earlier builds left on %s\n", n, h.on)
+	}
+	return n
+}
+
 // returned records that the build's work ended, and says whether a kill came first — only then is
 // the build killed; an error it ended with on its own is its own outcome.
 func (h *holder) returned(r *running) bool {