Spark init log growing in size over time

The spark init file is growing in size. I've found the cause to be that the call to `sparkthor.sh start` and the subsequent pid for the java process can't have 'wait' called upon it, due to it not technically being a child of the original shell. I've created a small wait_external that does a simple poll with a sleep, which is more efficient than the infinite loop and check that is occurring.