B02002 - drawbridge vs longworm

If a long worm's head is on the drawbridge and a tail segment is at the
portcullis and you raise the drawbridge, bad monster handling occurs
because of some recursion that occurs before set_entity is called again.
Not sure when this last worked; it's broken in 3.4.2 as well.  Modified
e_died to ensure both entity objects for the same monster get cleaned up
so subsequent e_at calls behave as expected.

doc/fixes34.3

@@ -95,6 +95,7 @@ resuming interrupted 'A' sometimes ended with "You finished disrobing" twice
 when you're asleep you shouldn't "notice" monsters that have become undetected
 must be able to reach floor in order to use stethoscope on corpse or statue
 fix a few coordinate (y,y) -> (x,y) typos in apply.c, mon.c, and wizard.c
+killing a long worm on a drawbridge could produce a panic
 
 
 Platform- and/or Interface-Specific Fixes

src/dbridge.c

@@ -399,6 +399,8 @@ int dest, how;
 		/* we might have crawled out of the moat to survive */
 		etmp->ex = u.ux,  etmp->ey = u.uy;
 	} else {
+		int entitycnt;
+
 		killer.name[0] = 0;
 		/* fake "digested to death" damage-type suppresses corpse */
 #define mk_message(dest) ((dest & 1) ? "" : (char *)0)
@@ -409,6 +411,13 @@ int dest, how;
 		else		/* you caused it */
 		    xkilled(etmp->emon, dest);
 		etmp->edata = (struct permonst *)0;
+
+		/* dead long worm handling */
+		for (entitycnt = 0; entitycnt < ENTITIES; entitycnt++) {
+		    if (etmp != &(occupants[entitycnt]) &&
+			etmp->emon == occupants[entitycnt].emon)
+			occupants[entitycnt].edata = (struct permonst *)0;
+		}
 #undef mk_message
 #undef mk_corpse
 	}