Thank you for your responses.
Radek, to explain more about reasoning behind the need of naming phases: imagine you are having several slighty different xmls with different configuration and phases. Starting to solve a problem you are able to identify the scenario and use right xml file with right phases. In each configuration file you have several phases, because - for example - you want to implement different algorithm for first half of local search (to find most promising direction of wide possibilities) and then another one for the ending time (to search optimal local solution).
However, at some point, everything will become messy to oversee. You can have trust in your methodology and files and phases you use, but you fail to actually remember every detail and logs are suddenly useless as they don't answer most fundamental question - what exactly just have run?
So, to give example of such names, in my case it would might be something like: "InitialConstructionHeuresticPhase", "GreedyTabuSearchPhase", "BasicTabuSearchPhase" "StrictTabuSearchPhase", "FinalCorrectionsPhase"... And so on and so on. There are just so many combinations of possible algorithms and approaches to be combined in optaplanner, but in the end just from looking at logs you cannot determine which combination have run - instead you have to debug which xml file was fired and then read the phases from the right file.
I hope this explains my need more.
Lukáš, I would love to add feature request - what would be the best way to do it properly in your opinion? Where should I write it?