Add metric estimation of page fault time

[svalat]

Would be nice to estimate page fault time.

When the tool start we can make a time measurement of page faults from sequential to parallel.

Then we can intercept mmap and count new pages max loaded into the app (will be an estimation because we are not sure it touch it).

Then just multiply by min/max page fault time. We need min/max because we know actually linux kernel page fault don't scale so if they are made in parallel this will kill the perf with high factor.

Need to take care in the GUI to well show this is an estimation and not an explicitly measured cost.

[svalat]

Could also make a time chart to see estimated % over time.

Ideally need to take in account the non scalability of the linux kernel, need to think how to compute estimated min/max.