On your point about Everything is happening in two dimensions, I advocate for a point-crawl structure: its quick to sketch on the fly, it boils the geography down to the essential details (what links to what) without overwhelming everyone, and it allows for some tactical complexity to build up.
For example, lets take something you might have written off as too complex for theatre of mind in your blog: The PCs face off against a half-dragon cult leader, his four kobold guards (two with bows) and their spider swarm pets.
I would sketch out something like the below for my players:
and say something like:
“As you emerge into the pillared entrance hall you see the cultist on the far side of a narrow bridge before a towering statue of his insane god. Two kobolds with spears and pilfered mail block the bridge, while two more lurk in the shadows above him atop a rickety tower. From the dark chasm underneath the bridge the rustling and chittering of sharp mandibles can be heard, and they are growing louder”
I think this does a good job of conveying the fiction and avoiding nasty player surprises (“can we push forward two abreast down the bridge? Oh wait never mind, I guess not if its narrow, huh?”)
and in a context of ‘rulings not rules’ you have lots of fodder to generate new options as you need (“is there really no way down from the archer’s tower to the bridge? What if I tie my rope and swing down?” “…I’ll allow it, make a dex roll.”)
This setup breaks your rules 1,2, 4 and 5 in fact, and we could easily add in a blood-slick floor to the Far Hall to break rule 3 too. Do you still think it would be boring/hard/uninteresting to run a combat this way? Genuine question 