[mapguide-internals] rfc90 ready for review

[UV]

Hi,
I commited revision 6174 of the meta tiling code RFC90 in sandbox
http://svn.osgeo.org/mapguide/sandbox/rfc90.2/MgDev

The metatiling using lock files per metatiles is working fine.
Somehow the approach using ace_conditions seems to break. I suppose due 
to datastructures shared between threads.
I would appreciate if somebody cares to have a look at it, as I 
generally thought this approach to be a highly efficient way to 
implement the locking of the tile cache. When tracing this with the 
debugger or monitoring the logfiles
the ACE_CONDITION locks work as expected, however in a release version 
something starts to go wrong quickly.

As the file based locking works fine I think this is ready for review as 
the alternative locking strategie can be easily
removed.

I ran a few memory checkers on this code which did not find any further 
memory leaks.
However, a small increase over long running time can be observed as it 
can without metatiling.
Since the metatiling mode is only used in seeding tile caches and not 
for ongoing production I suppose this is tolerable behaviour.

Please comment.