address issue #1490 - rolling boulder segfault

Issue reported by BartekCupial:  segfault occurred and was tracked to
behavior of a rolling boulder trap.  Suggested fix was included, but it
assumes that the destination spot is valid so is suspect.

A comment pointed out that the path is validated when a rolling boulder
trap is created so the segfault should be impossible.  I didn't find an
explanation but am adding a fix based on the one in the issue report.

Closes #1490

doc/fixes3-7-0.txt

@@ -1602,6 +1602,7 @@ when starting a new game as a monk, newsym(0,0) was being called (adjabil ->
 	postadjabil -> see_monsters when initializing See_invisible as hero
 	became level 1; seen 'monster' was the hero who hadn't been placed
 	on the map yet)
+incorporate a fix to prevent segfault due to rolling boulder trap
 
 
 Fixes to 3.7.0-x General Problems Exposed Via git Repository

src/trap.c

@@ -3375,6 +3375,23 @@ launch_obj(
             while (tmp-- > 0)
                 nh_delay_output();
 
+        /*
+         * TEMPORARY?  github issue #1490 by BartekCupial reports a
+         * segfault when boulder rolls out of bounds.  That should be
+         * impossible because trap creation validates the path that
+         * the boulder will traverse.
+         *
+         * The suggested fix increments bhitpos, verifies with isok(),
+         * then undoes the increment if not ok.  This is simpler.
+         */
+        if (!isok(gb.bhitpos.x + dx, gb.bhitpos.y + dy)) {
+            x2 = x, y2 = y; /* use current spot for final boulder placement */
+            break;
+        }
+        /*
+         * end TEMPORARY?
+         */
+
         x = (gb.bhitpos.x += dx);
         y = (gb.bhitpos.y += dy);