Experiment with different cache replacement algorithms for the parser cache

[andresriancho]

We're now using the LRU algorithm to replace the items in the parser cache, but maybe it's not the best thing to do?

Check these cache replacement algorithms

• https://github.com/uqfoundation/klepto
• http://code.activestate.com/recipes/576532-adaptive-replacement-cache-in-python/

Run different tests using the new profiling features

This is the output from a random scan:

$ cat /tmp/w3af-*core | grep hit_rate
        "hit_rate": 0.26894865525672373, 
        "hit_rate": 0.26875, 
        "hit_rate": 0.29097605893186, 
        "hit_rate": 0.28468468468468466, 
$

~27% hit rate is really high for any cache!

[andresriancho]

Related with Effective use of memory/CPU #9068