when you don't know the timeline of the project you're working on. So you don't know if you should rewrite everything to be more flexible in the future

I generally end up over engineering everything so the code is general enough to accommodate any future features. I do this in every single project I work on. After a few hours of trying to come up with a good solution I realize I'm just wasting time, so I scrap everything and go with the simplest solution. Sometimes it's easier and faster to write simple code and if the requirements change, just rewrite it.

I usually try to predict if the client will want to also have B, C and maybe even D along with the A..
If the answer is yes, most probably I'll put in some time to rewrite everything as it'll be easier to implement B, C and D in the cleaner code without extra spagetti voodoo..
Of course, I explain & discuss this with the boss.. sometimes we go straight to rewriting, sometimes it's implement A right away so the client is not pissy, but start on rewriting and impelemnting A, B, C & D as soon as you finish with old A..

If you're in the position to make such decisions alone, check if things can easily be added/changed in the old code.. if not, I think it's worth rewriting for the long run benefits.