Fix race condition crashes when unmapping views

This fixes two issues which were both introduced in #2396.

First issue:
The PR changes the location of the buffer save to transaction_apply, but
puts it inside the should_configure block. For unmapping (destroying)
views, should_configure returns false so it wasn't saving the buffer. If
a frame was rendered between the unmap and the transaction applying then
it would result in a crash.

Second issue:
If a destroying view is involved in two transactions, we must not
release the buffer between the transactions because there is no live
buffer to grab any more.

sway/desktop/transaction.c

@@ -195,6 +195,12 @@ static void transaction_apply(struct sway_transaction *transaction) {
 				sizeof(struct sway_container_state));
 
 		if (container->type == C_VIEW) {
+			if (container->destroying) {
+				if (container->instructions->length == 1 &&
+						container->sway_view->saved_buffer) {
+					view_remove_saved_buffer(container->sway_view);
+				}
+			} else {
 				if (container->sway_view->saved_buffer) {
 					view_remove_saved_buffer(container->sway_view);
 				}
@@ -203,6 +209,7 @@ static void transaction_apply(struct sway_transaction *transaction) {
 				}
 			}
 		}
+	}
 }
 
 static void transaction_progress_queue() {
@@ -276,9 +283,9 @@ static void transaction_commit(struct sway_transaction *transaction) {
 			// mapping and its default geometry doesn't intersect an output.
 			struct timespec when;
 			wlr_surface_send_frame_done(con->sway_view->surface, &when);
-			if (!con->sway_view->saved_buffer) {
-				view_save_buffer(con->sway_view);
 		}
+		if (con->type == C_VIEW && !con->sway_view->saved_buffer) {
+			view_save_buffer(con->sway_view);
 		}
 		list_add(con->instructions, instruction);
 	}