gfgtdf

The reason for the previous PR was to make the code faster in particular by making it easier for the compiler to use simd instructions. Making the code more complicated...

> It definitely shouldn't use "cumulative" in the name or as the main description - we already have to many different meanings attached to that word. What is it expected...

One of the problems for me is that in case something goes wrong and we forgot to catch something we currently catch (and show in the errolog) an instance of...

im actually wondering whether we still need lexical_cast(_default) in the first place, while its kinda cool to be able to convert beween arbitiary types its imo not something we actually...

> I wanted to replace utils::from_chars with lexical_cast hmm i guess that woudl work too, a disadvantage of that approch is though that a generic lexical_cast doesn't allow additional parameters...

i'm not sure whether that works, in particular because replays can start at an aribitiary point, if for example the [change_theme] happens in turn 1, and then the user wants...

alternativeley we could add the default replay REPLAY_THEME to all themes by default, by having some `[default_replay_theme]{REPLAY_THEME}[/default_replay_theme]` in the game config and then falling back to that if it doesn't...

as i see it this disables enemy checking filters checking in abiliies during ai combat simulations. i'd prefer not to do it since enabling these filter filters was one of...

i don't really see the problem with the previous solution

yes that seems likely, if someone has the time, it woudl be nice if anyone could checke whether it still hppens with latest 1.18 for example https://github.com/wesnoth/wesnoth/actions/runs/11193473424/artifacts/2019152310