You can already undo up to 5 times by default, but you can up that. Do note though, it's so low by default because snapshots (the history entries required for undo and redo to work) are massive and a high enough undo limit can cause TPT to run out of memory. This can happen even if your system reports low memory usage, if it limits how much memory TPT is allowed to allocate. TPT does not go out of its way to ask for this limit to be increased.

As for "fixing" this, sadly, undo history in TPT is already as good as it gets. We can't just do what text editors do, which is to store the actions the user takes, effectively storing a copy of the text in a chain of undo history items, which overall aren't too much bigger than (i.e. some constant multiple in size of) the actual text in the buffer. This is because while in text editors, the only actor changing the data (the text) is the user, in TPT, there is also all the simulation code changing the data (the simulation). Thus, the history is actually a combination of the user's actions and the simulation code's actions. The latter would be even more expensive to store than what we do store right now, which is just differences in snapshots of the simulation taken when the user does anything.