Neither freeing slabs nor purging the cache are safe to call from interrupts, so instead we schedule these to run on NetSurf's process.
If NetSurf is busy, there may be significant delay, so the priority is set low to ensure this is called only when other - faster - memory handlers have been exhausted.
I don't know if this works, or is sane, as I'm not entirely sure how to test it.
Something similar needs adding for OS4 but low memory is less of an issue there, so will see if this works on OS3 first.
