* improve profile performance
* support profile nested class
* support total_wasted time
* calculate totalChildrenWasted time for echarts
* only profile super class in the same package
* improve perf through local-tree,local-timer,local-stack
* improve test case
* add Stopwatch remove-child test
* fix can't profile parent class with 2 class not in the same package
* add totalChildrenTimes
* improve LocalTimer perf
* add PerfUtil.useLightStopwatch(true)
* define behavior of switching light-stopwatch process
* ignore other threads if profileSingleThread(true) with multi-threads
* improve the empty judgment the first call in useLightStopwatch()

Change-Id: Id1ae075ddffec77b95b31142d7ebcdae87371943