I fixed the remaining major problem in the tourney manager, of the hanging processes. Unlike what I thought first, this had nothing to do with machine workload or terminating processes (although the latter remains a problem area in WinBoard), but was an ordinary bug. I also fixed the incompatibility of the -fSAN -sSAN options with walking the PV, which I discoveredin Leiden.
I pushed thepatches to hgm.nubati, and also fast-forwarded hgmaster along the experimental branches that originally were created to improve the position search, but in the mean time had accumulated lots of other stuff that was not easy to commute to the beginning of it. As I have been using all that stuff now for some time already as my main version, it can be safely added the development branch. This would then be the version that should become 4.6.0, and has no loose ends I am aware of..
