Enable dynamic pair list pruning
This change activates the dynamic pruning scheme and the pruning
only kernels added in previous commits.
A heuristic estimate is used to select value for nstlist and
nstlistPrune that should result in performance that is reasonably
close to optimal. The nstlist increase code has been moved from
runner.cpp to nbnxn_tuning.cpp. The KNL check in that code has been
replaced by a check for Xeon Phi.
A paragraph has been added to the manual to describe the dynamic
and rolling list pruning scheme. A reference with all the details
will be added once the paper has been published.
Change-Id: Ic625858a07083916c8aa3e07f7497488dcfaee9e