Fix ticks wrap issue of lowmem test in vm_pageout_scan

[smh]

When looking at this again, I couldn't see the issue you mentioned in IRC last night John:
[21:24] <BigSpoon> it means it only does it
[21:24] <BigSpoon> if you scan within the window
[21:24] <BigSpoon> it should be >
[21:24] <BigSpoon> not <

Have I not had enough coffee this morning?

The overflow issue was there, which I've addressed in this revision.

[jhb]

I had it backwards, the condition is correct. However, this is not the correct way to handle wrapping. You should always compute a tick delta and compare that instead. That is:

ticks - lowmem_ticks >= lowmem_period * hz

Will would prefer you to do a divide to avoid overflow, so that would become:

(ticks - lowmem_ticks) / hz >= lowmem_period

I think this is also easier to read and parse than the current expression (is the amount of time that has passed greater than that the throttle period?)

[smh]

Yep got confused looking for the range issue, updated with the recommended method for signed vals.