gh-47: update wal/oom.test

Tuples are got a big bigger with shared-arena patch (4 bytes
per tuple overhead).

So we exhaust the current arena in out-of-memory test a bit
faster than before.

test/wal/oom.result

@@ -4,7 +4,7 @@ error: 'Failed to allocate 19436 bytes in slab allocator for tuple'
 ...
 lua box.space[0]:len()
 ---
- - 4853
+ - 4852
 ...
 lua i = 1 while true do box.insert(0, box.space[0]:len(), string.rep('test', i)) i = i + 1 end
 ---
@@ -12,7 +12,7 @@ error: 'Failed to allocate 4095 bytes in slab allocator for tuple'
 ...
 lua box.space[0]:len()
 ---
- - 5871
+ - 5869
 ...
 lua i = 1 while true do box.insert(0, box.space[0]:len(), string.rep('test', i)) i = i + 1 end
 ---
@@ -20,7 +20,7 @@ error: 'Failed to allocate 2051 bytes in slab allocator for tuple'
 ...
 lua box.space[0]:len()
 ---
- - 6378
+ - 6375
 ...
 lua box.space[0]:select(0, 0)
 ---