After reducing the number of operations in the initial_sync_wt_cache_full.js (see SERVER-36969) from 1000 to 500, there has been one documented instance of a similar timeout failure on a slow machine. We should be able to lower the number of operations further to reduce the likelihood of the test timing out.