staleness visualization performance too slow: only compute for snapshots shown in graph

_From [derek.br...@gmail.com](https://code.google.com/u/117968039472581148324/) on December 10, 2010 19:52:23_

PR 611338

from Bharath:

staleness has one important open issue -
large number of snapshots with staleness visualization is
quite slow - the problem is that I am computing staleness for all
snapshots instead of the ones shown in the graph - need to build an index
for staleness.log (just like the other ones) and compute staleness only
for those snapshots that are visualized

from Bharath's notes:
-1. staleness with 10k snapshot is really slow at startup - more than 5 min!
note: dump snapshot.idx and callstack.idx from the client & process only
those entries that are needed - processing all entries is what is
causing problems
if snapshot numbers are reset to zero after a nudge then this might be
a problem! ditto for callstack numbers too

_Original issue: http://code.google.com/p/drmemory/issues/detail?id=233_