turn-overflow on blocked monsters

screenshot to reproduce

Logged In: YES
user_id=990499

Of course it is non-reproducable :-/

But i get a few overflows which are not created by endTurn()
The last number is the number of milliseconds used for the
complete endTurn() method.

- 0 3 3 3 3
- 0 4 4 4 4
- Turn duration overflow by 66 ms
- 0 4 4 4 4
- 0 3 3 3 3
- 0 4 4 4 4

Logged In: YES
user_id=990499

After profiling marauroa.server.game.RPServerManager: The
problem is in beginTurn();

252877 - 0 0 0 0 4 4 5 8 8 8 279 279
253188 - 0 0 0 0 3 3 5 9 9 9 285 285
253477 - 0 0 0 0 3 3 5 9 9 9 270 270
253786 - 0 0 0 0 3 3 5 8 8 8 276 276
254101 - 0 0 0 0 4 4 6 9 9 9 287 287
254435 - 0 0 0 0 4 4 6 8 8 8 317 317
254435 - Turn duration overflow by 17 ms
254993 - 0 0 0 0 3 3 5 9 9 9 558 558
254993 - Turn duration overflow by 257 ms
255291 - 0 0 0 0 3 3 5 8 8 8 297 298
255604 - 0 0 0 0 3 3 17 19 19 19 306 306
255605 - Turn duration overflow by 6 ms
255892 - 0 0 0 0 3 3 7 10 10 10 287 287

endTime[9] = System.currentTimeMillis();
ruleProcessor.beginTurn();
endTime[10] = System.currentTimeMillis();

Logged In: YES
user_id=990499

Oh, it is creature.logic(). The reasion why it is in
beginTurn() instead of endTurn() is that summoned monsters
are treated as NPCs. (They are in the npc-collection because
they have no respawn-point.)

Logged In: YES
user_id=990499

First guess was right: The time is spent in pathfinding
until a timeout of 100ms is reached. Average pathfinding
time is 0.095 ms (after spending some time in Nalwor).

I decreased it to 10ms because 100 times the avarage value
should be more than enough time to find a suitable path.
There are a few cases in with a path is found in 100ms but
not in 10ms but this can be ignored. In the cases i found
the creature is nearly 2 screen-sizes away and needs to walk
further away at the start of the path.