I added a comment to issue207 (revise search code) that the output for lazy
search should be clearer. Since LAMA/FF synergy behaves correct in this respect,
I close this issue.

Just to make this clear: This is really only an output problem, the heuristic
estimate of the goal state is 0.

This does not really look like a bug. The behaviour is triggered by the
following code lines in the lazy search:

            if (check_goal_and_set_plan(current_state))
                return SOLVED;
            if (search_progress.check_h_progress(current_g)) {
                reward_progress();
            }

The output of the "Best heuristic value" is a side effect of
search_progress.check_h_progress, which is not called for goal nodes. Should we
change this? I am not sure whether it is worth to make the code more complicated
for this, especially since we are also planning a major revision of the search
implementation anyway.

It would be interesting to see whether the problem still exists with the recent
changes to the landmark code. Salomé, could you please try this?

When running the task from Issue127 with the LAMA/FF synergy, the best heuristic
value does not drop down to 0 before finding a goal state, even when we are not
using reasonable orderings. Use e.g. the following call:

 ./downward --heuristic
"hlm,hff=lm_ff_syn(lm_rhw(reasonable_orders=false,cost_type=2,lm_cost_type=2))"
--search "lazy_greedy(hlm, hff, preferred=(hlm, hff))" < output