The first thing that needs to be addressed is the preexisting divide of Tsukihime and Fate as had already ossified in fandom consciousness, to wit the allocation of focus on the Church/vampire conflict and magi respectively. This was patently wrong: Fate was and always will be about Servants, with every work released under that label subordinated to that single core thematic. This was not readily apparent with only a small sample size of available works which in operating as building blocks of a unified cosmology did not yet challenge the notion of a logically consistent shared setting, but became so when the illusion of balance between the elements of this conceived shared framework of reference that was the "Nasuverse" could no longer accommodate the reality of how skewed the narrative focus of works notionally placed in that setting had become. It was no longer a matter of timelines not aligning perfectly or two Shikis unable to live under the same sky because the odds didn't work out when the moon is a supercomputer and it's a common occurrence for any random magus to set up a ritual anywhere in the world which coughs up seven or more superghosts that barely anyone can even scratch. Superghosts and supervampires being mutually exclusive is simply a codification of the simple fact that the over-development of one notional side of the coin has outstripped not just that of the other side but the capacity of the coin itself.
What that underdeveloped other side of the coin really was can no longer be said, as it never had the chance to grow beyond its initial form and context. It is easy to say that magi "take a back seat" in Tsukihime and Kagetsu Tohya because none of the characters happen to have that be their foremost identifying quality, and similarly it can be said that Tsukihime as a work is marked by narrative interiority, predominance of the psychological element, and, of course, a self-contained, grounded, "small-scale" plot. But what must be considered is firstly that Tsukihime was written without the benefit of an existing foundation for the setting in which it took place - and consequently lacking a thematic basis for readers to be familiar with and have preconceived notions on - and so had to construct that notion by laying the groundwork for this cosmology - the rules - while providing peeks above the immediate presentation and behind the proverbial curtain to a perceived depth beyond what is seen in the work itself - the lore - while also writing a self-standing story which could become a commercial success. What we get from it is the result of trying to balance these factors in the writing process, and in that also the defining features of Nasu's own writing. Which is to say, the essential information required for the story to acquire a "mythological" background and for its events to be consistent with it - the "rules" of the work that the reader is asked to keep in mind while reading. This is nothing to characterise an entire hypothetical series of works set in the same framework by. Lest we forget, Tsukihime II was the ensemble cast that all ensemble casts ever conceived both in fanfiction and official works emulated. A clusterfuck of magi and vampires and churchies coming to a head with the most powerful characters engaging in high-octane large-scale action over world-ending stakes is about as far a sequel could get from the aforementioned description of the first installment.
Then there is the matter of the notional pie, where a slice can have any proportion of toppings of the one or other kind. There's two ways to approach the simile:
A storytelling perspective: the model of a unified cosmology is unsustainable insofar as the illusion of consistency in the "rules" that govern the reader's conception of a notional setting has been constructed and elevated in importance by the author and his constant snipping and trimming over the years to keep the pointy ends from sticking out of the uniformity and getting in each other's way to the point that it has become a decisive factor of not just quality but legitimacy of a given work. As the works multiply and deviate the capacity of the model to contain them all is strained beyond the point where keeping to the rules makes a parody of either them or the work itself. Dividing and keeping like with like is thus simpler, cleaner, easier; that even this partition could not contain the deviation, necessitating even more extreme rationalisations through amendments to the cosmology is a separate matter.
A business perspective: people like the Servant topping. The Servant topping sells a lot. Pie doesn't sell a lot. We should give people more Servant toppings, but in food that sells more. What, the vampire topping? We promised to bring it back in a pie, but pies don't sell a lot. But we can't put it in another kind of food before we bring back the pie. Besides, people can't get enough of the Servant toppings. Try putting in just a little bit of the vampire topping along with the Servant toppings and see if people like it in other foods.
Now I'm hungry.