-P option now sets target P_N, not P_1. Bumped to version 0.9.

-P option now sets target P_N, not P_1.  Bumped to version 0.9.

This cleans up the trouble I had before with
  testing/bell_rate/bell_rate.sh 50 1e-6 0.05 1e-3 1e-9 300
where the large number of domains was making P_N much larger than P_1.
This lead to more "ignored double unfoldings", creating a banded
structure in the simulated histogram.

The new '-P sets target P_N' configuration makes it more
straightforward to use -P for it's intended purpose, avoiding "ignored
double unfoldings", without forcing unnecessarily long timesteps once
the population of the limiting state is reduced.

Note that the second purpose of the -P option, avoiding significant
force changes in a single timestep, is either unaffected (for states
with population 1) or improved (for states with population > 1), but
this should perhaps be separately controllable through it's own
"--dF-max" option, or similar...