Update stat documentation to reflect allocation strategy.

The documentation stated that memory allocation was always in chunks
of 1MB.  In practice, this gets less effecient as you get into larger
slabs.

The 1MB thing (which can be controlled by -I) is both the maximum size
and allocation unit so a single allocate event will attempt to get as
many items for a particular slab as it can as long as the allocation
itself doesn't exceed the maximum item size.

doc/protocol.txt

@@ -542,8 +542,8 @@ END\r\n
 | chunk_size      | The amount of space each chunk uses. One item will use   |
 |                 | one chunk of the appropriate size.                       |
 | chunks_per_page | How many chunks exist within one page. A page by         |
-|                 | default is one megabyte in size. Slabs are allocated per |
-|                 | page, then broken into chunks.                           |
+|                 | default is less than or equal to one megabyte in size.   |
+|                 | Slabs are allocated by page, then broken into chunks.    |
 | total_pages     | Total number of pages allocated to the slab class.       |
 | total_chunks    | Total number of chunks allocated to the slab class.      |
 | get_hits        | Total number of get requests serviced by this class.     |
@@ -565,9 +565,9 @@ END\r\n
 
 * Items are stored in a slab that is the same size or larger than the
   item.  mem_requested shows the size of all items within a
-  slab. (total_pages * 1MB) - mem_requested shows memory wasted in a
-  slab class.  If you see a lot of waste, consider tuning the slab
-  factor.
+  slab. (total_chunks * chunk_size) - mem_requested shows memory
+  wasted in a slab class.  If you see a lot of waste, consider tuning
+  the slab factor.
 
 Other commands
 --------------