Done

Ah, sorry, I should have read the old messages. I think I still find this a bit
confusing...

I think it would help to make the messages more verbose by mentioning the g and
h value for the "new best" expanded state in the "Best heuristic" line. Then it
would be clear to the reader that they belong to a different f layer. Can you
add that?

The reported numbers are correct (and always have been as far as I know).
It's just that the check for best-heuristic-value is done when a state is
generated, and for f-value when a state is expanded.

Are they messages clear about what they refer to? Down in the first comment, it says
    Best heuristic value: 6 [expanded 103753 state(s)]
which would be a terrible message if it referred to generated and not expanded
states!

The f-message still refer to expanded states and the best-heuristic-value
message still refer to generated states, but this is just because these are the
places where these things matter.
The output looks the same now (also between different searches).
Malte - I guess it's up to you. If you want to do something else (maybe don't
output best-heuristic-value message in A*)?

Erez, would you say this is done?

Added a SearchProgress class which is used by all searches

The progress statistics in the eager search algorithm are currently a bit
confusing because some of them refer to *expanded* states, others to *generated*
states. For example, consider this fragment of output in the Elevators domain
(option "ou"):

f = 47 [551962 evaluated, 52462 expanded, 64 reopened, t=159.9s]
f = 48 [768713 evaluated, 76110 expanded, 64 reopened, t=222.63s]
f = 49 [1018454 evaluated, 103685 expanded, 64 reopened, t=295.17s]
Best heuristic value: 6 [expanded 103753 state(s)]
f = 50 [1362992 evaluated, 138060 expanded, 80 reopened, t=393.26s]
f = 51 [1916551 evaluated, 192159 expanded, 84 reopened, t=553.87s]
Best heuristic value: 0 [expanded 192172 state(s)]
f = 52 [2571654 evaluated, 255295 expanded, 116 reopened, t=742.23s]

The "f = ..." output lines correspond to expanded states, but the "Best
heuristic value: ..." lines correspond to generated states. That's somewhat
confusing, as it would be natural to assume that the lines after something like
"f = 51" correspond to states that belong *to that f layer*, which is not the case.

Suggested change: Only create output for *expanded* states, not generated states.