simpleperf: Add CallChainJoiner.

Add CallChainJoiner, which uses LRUCache to cache call chains in memory,
and extend a call chain if the (tid, ip, sp) tuple of its top nodes appears
in the cache.
It is added to break the 64K stack limit when recording dwarf based call graph.

Bug: http://b/69383534
Test: run simpleperf_unit_test.

Change-Id: I70f51e13c1e312170be6468ee6962e5e9a48faef
4 files changed