I've complained in the Stellaris thread about how they introduce tile and building mechanics in the early game to give the player something to do, then need to automate it away once the player's empire reaches a certain size. It is like having one set of feature programmers dig a hole that the AI programmer then has to fill in. Adopt more abstract and scaleable mechanics from the start and you save a lot of player and developer time!
Someone made the point on the Hearts of Iron IV forum that game has gone from building corps, to building divisions to specifying the division composition at the brigade level to specifying composition at the battalion level. Each step has amounted to a larger and larger player bonus as the AI becomes less and less competent at building an effective force and responding to player innovations.
I am starting to think it is very difficult to successfully adopt detailed mechanics that don't operate at the main level of the game. Each "game within the game" that you have tends to end up increasing player workload. If the work is interesting, then it risks giving the player an advantage that can trivialize the AI, and if it isn't interesting then it's meaningless. If you make it interesting AND a competent AI, then you still have to avoid forcing the player to do it too often and avoid having the "game within the game" break the strategic level in subtler ways.
Yet, I still love ship design and tactical combat in a space 4x. Cold dead hands, etc.