You wouldn’t normally use a limit switch exactly for a z axis zero, although it’s essentially the same thing. Usually you have some very thin conductive material that sits under the bit on the surface that you want to zero to, and then you have a wire that is grounded to the bit as well. Lower the z axis until the circuit completes. So, works pretty much just like a limit switch electronically, but it’s removed after the zero is set so that it isn’t in the way.
As for the board itself, if you do design it all on one board, I’d suggest making sure you have it designed in a way that the controller portion is easily modified and changed to a different controller. There are threads where people have been exploring porting the Maslow firmware to more powerful controllers. If you were to design this board and produce it in mass quantities it’d be a shame if a better controller option is then found and there is no easy way for you to modify it. I think this is one reason that it makes for a very good argument to keep the controller and power board separate.
I’d also suggest not making it to small. I think it’s a good idea to keep a bit of space between the H-Bridge chips, since those are big heat producers.
One other thing you might consider is having possibly vacant but easily added places for additional H-Bridge chips / motor connections. There has been talking and possibly some people playing around with adding additional motors, possibly up to 5 I believe (one for each corner and z-axis). If your board could readily be adapted to accommodate this that would be a bonus IMO. Heck if you wanted to go ahead and include the H-Bridge chips and motor connections, even if using more motors doesn’t become standard, you could design in some way to specify which motor the H-Bridge is for so that it could be failover chips if nothing else. Burn up one chip and move to another without having to replace the board hehe Don’t know if the additional cost would be worth it though.