Yes, right about the cleanup handler.

Your Apache::Reload idea sounds fine, but the code to stat a file and run your config parser again is awfully simple so don't be afraid to just write it.

I don't think there's much difference between the config modules in terms of reliability (it's a fairly simple task), but for determining memory footprint I think you will just have to load them into memory to see. Larger files usually lead to more memory, unless the file is mostly POD.

Don't worry about file locking. When you want to update your config file, just do it with a rename, i.e. mv myconfig.conf.new myconfig.conf. That should be an atomic update.