Checker reports incorrect scores for induced cost heuristic with k-clustering

[schrins]

Similar to the recoded heuristic not producing consistent results, the old implementation has issues when using k-clustering. I used the same data as described in this issue:
#21

In addition, I used the following parameters to produce the following results:
-v 3 -r 000000 -c -H -k 4

100: Could not verify the editing costs ... found :817.71 while it should be 819.858
Verifying that solution 0 is a valid solution for WCE ...
Solution validity verified!
200: Could not verify the editing costs ... found :3821.19 while it should be 4160.22
Verifying that solution 0 is a valid solution for WCE ...
Solution validity verified!
300: Could not verify the editing costs ... found :8655.1 while it should be 9262.46
Verifying that solution 0 is a valid solution for WCE ...
Solution validity verified!
400: Could not verify the editing costs ... found :13746 while it should be 13922.8
Verifying that solution 0 is a valid solution for WCE ...
Solution validity verified!
500: Could not verify the editing costs ... found :18772.2 while it should be 23551.6
Verifying that solution 0 is a valid solution for WCE ...
Solution validity verified!
600: Could not verify the editing costs ... found :24973.4 while it should be 30590.8
Verifying that solution 0 is a valid solution for WCE ...
Solution validity verified!

Edit:
If I use the samples without sequencing errors, the issues are even more extreme:
100: Could not verify the editing costs ... found :0 while it should be inf
Verifying that solution 0 is a valid solution for WCE ...
Solution validity verified!
200: Could not verify the editing costs ... found :1347.57 while it should be inf
Verifying that solution 0 is a valid solution for WCE ...
Solution validity verified!
400: Could not verify the editing costs ... found :4796.57 while it should be inf
Verifying that solution 0 is a valid solution for WCE ...
Solution validity verified!

I do not know where the "inf" comes from. It would mean that we put a forbidden edge into a cluster or separated a permanent edge.

[phspo]

should be closed now?