Memory Leak

[KatieWoe]

Due to phetsims/atomic-interactions#72 it was decided a short memory test should be done on the recently published sim. Results:
Start: 27.2 MB
1 min: 56.6 MB
2 min: 74.9 MB
3 min: 94.2 MB
4 min: 119 MB
5 min: 137 MB

[jbphet]

I tested a RequireJS version (1.1 release branch) with the fixes in the commit above, and the results look good. Here's a table:

Time (minutes)	Memory Usage (MB)
0	30.2
1	37.5
2	39.3
3	40.3
4	40.7
5	41.0
6	41.2
7	41.4
8	41.7
9	42.0
10	42.2

The memory is increasing, but not by much, and the total is well within the expected range for a sim of this size and complexity. It's also more than a 3x reduction in memory usage at the 5 minutes mark. And, finally, I looked at several comparisons of the heap snapshots captured during the testing, and there were no leaks of SOM-specific objects. All increases in memory usage were from things like "array" and "compiled code". Here is a screenshot from a comparison of two snapshots:

The bottom line on this is that I think the leak is fixed, and I should do a maintenance release on it.

[jbphet]

I should do a maintenance release on States of Matter 1.1 to publish this fix.

[jbphet]

Redeployment complete. I tested the built version before deployment, and the memory numbers looked good (a little better than in #237 (comment) actually). Closing.