How can you have an expansion phase in board game design considering it's already constrained in time and space?

As a non-game designer of the industrial variety I think that you're talking about two things.
First the expansion-compresion phases are what's called, in Design Thinking terms, the double diamond process, where there is a divergence phase, where you do your research, brainstorm ideas and in general broaden your perspective on the project and possible solutions, then you fall into a convergence phase where you take everything from the first phase and turn it into an actual product (be it an app, a toaster or a boardgame). Once you advance with the project you'll start testing your solutions, so you'll have another divergent phase that, again, will lead to another convergence phase where you'll take everything you've learned from testing and apply it to your product.
This is will happen several times during the design process as more and more iterations happens. Now, that's just a normal design process but I see you talking about another thing that I believe that's just mixed up on this. You talk about videogames being more about adding and boardgames about removing. I don't think that's related to the process rather than the constrains of the medium. With videogames the computer can take a lot of the mental load off the player so you can create depth by adding complexity. Board games, on the other hand, depend on the player's mental capacity supported by just cardboard and plastic bits. In this case you can't generate depth by adding complexity, you generate it by designing systems with fewer elements but richer relationships, which of course involve a lot of streamlining.
Now, your main question, which is quite interesting, is "How an expansion phase could be possible in board game design considering you are always facing the restraints of reality."
To which my reply is that you'll always have a divergent phase in a design process, only you'll be looking at the problem from a different perspective. I propose that you CAN go wild with board game design ideas, you can explore without having to worry about the components of the game. Obviously the convergent phase will be different than with videogames because you're working with different constrains.
I suppose you're right that boardgames have much shorter iterative process than videogames* but I see it as a great advantage as you can adjust the design much much faster and get to something that's great quicker than with something that requieres more production time.

I think that themain issue you reflect on your second post has to do with the different kind of depth you can achieve with both mediums. Yes, videogame allows you to manage more information on the background but if designed well boardgames can be better at creating meaningful decisions and thus much more involved play.

Obviously I absolutely disagree that the "compression phase is where the """""real""""" design happens". Both phases are part of the process, you can't have convergence without divergence first.
Divergence is not about just listing ideas, it's about learning. If you don't learn about approaches, about mechanics, about components, if you fail to learn from playtests then your game will be crap, plain and simple. You can't design in a vaccuum.

When I was working on The Singularity Trap game (server/browser) I began with the constraints of the currently "minimum standard" computer/laptop display screen resolution of 1280x720 pixels and screen size 10+ inches to determine my limits on board size and then game token size since I wanted the entire gameboard to be visible all at the same time on the screen without needing to zoom in or out during play by players with normal visual capabilities.

It was this pixel limitation (and the just barely possible need to have 6 players all with a token in the same space on the board) that I had to deal with first. That and the whole 3d thing, which in the physical reduced hidden info version Fleet Admiral - Singularity is done with an altitude chit held in a fleet clip but in the server/browser version is done with a synopsis of each levels known content to the right of the board and an strategic (top down) view plus individual layer views.

After that display limitations part was resolved I began to build the database and the area map interactions etc and the commands interface and the individual unit data reports.

Going back to physical, with all fleet locations visible to all, introduced interception by other moving fleets during the move phase and sensors moved to providing ship and fleet details instead of locations in range, which also reduced backstabbing by allies somewhat. Optional tit-for-tat and averaging methods to eliminate excessive combat die rolls in the physical version.

But the basic point is that the digital game began by dealing with the limitations imposed by the available medium before it began any expansive phase.

From what you say, the divergent phase of board game would not only be thematic exploration but also mechanic searching which is much more important than in video game design which does not necessarily require new mechanics

I guess that the easiest way to think about it is to see it as two different mental states.
Divergence is an exploratory state, you're much more open to new ideas and influences, it's about what it can be. On the other hand, a convergence state is all about implementation. You're in a focused state. Your goal is to get things done. That doesn't mean creativity is not involved, we all know you need to be really creative to solve many of the challenges in game design, but the cute idea is that you're in a much more goal oriented mindset.