This isn't just a narrative problem, it's a physical IRL time problem. By level 7, nearly every PC has multiple things to do on every turn. Resolving 5 PCs and 5 Mobs turns can easily take 7 to 10 minutes. So completing even 18 rounds of combat is going to require 2 to 3 hours of IRL time combat. And that's if we're doing things fast.
But things can go really slow. Just look at a Monk burning some ki points.
- 2 Attacks, each with a Stunning Strike
- Flurry of blows for two bonus attacks and Way of the Open Hand forcing saves
Resolving such a Monk round requires up to 8 sequential rolls of the d20 (attack/save, attack/save, ...). What's worse, each roll depends on the roll before it. You can't roll your second attack until you know if the Stun worked. Your second Flurry attack may be improved if the target fails, etc.
There's just no way to resolve such a round in an IRL minute. So completing 18 rounds of combat requires hours of combat and sucks up entire sessions.
This is the "damage per second" problem that bogged down high level 4e combat. Your damage
per round could keep up or diverge, but quite often this came at the sacrifice of damage per second - table time it took to resolve something.
Being
brutal about damage per second when designing features and the combat engine is important, and I think 5e does run into issues with it.
But some of your examples aren't great. You
can roll a mass of attack rolls; so long as you know the order, you can apply the stun and retroactively add advantage.
Another trick that 4e tried (and failed at) was to avoid multi-tap attacks, and if you do do multi-tap attacks make them mechanically simple. 4e failed at this because multi-tap attacks ended up doing much more damage than single big attacks, but at least they tried.
---
If I took the "damage per second must remain high" rule and applied it to 5e, we'd do away with extra attack and replace it with more damage dice on a hit.
---
Combat Mastery:
Starting at level 5, at the start of your turn roll 1d20. This roll can be used to replace a weapon attack roll before the start of your next turn (you can do this after you know your attack missed). In addition, your weapon attacks deal two sets of weapon damage dice. In addition, if your attack reduces a creature to 0 HP, you can make 1 additional weapon attack this turn (if you aren't incapacitated).
---
I think this generates a similar level of damage per second to 5e "extra attack", but I tried to speed it up a bit. I avoided up-front decisions.
We roll 1d20 "mastery" die at the start of your turn. If you roll a 20, you know you are getting a crit on demand, which is fun. There is no decision here.
We then make an attack. This attack is not quite twice as powerful as a baseline 5e two attack routine, but because it is one attack the resolution is faster.
If it does drop someone, you get a free cleave. I put this here because it (a) doesn't happen often except against mooks, and (b) keeps needless decisions out of the primary execution path. I could pick "you can split your attack into 2 targets for 1 set of damage dice each" or something, but that would require an on the fly tactical decision each turn, and that slows play.
The fighter
Improved Combat Mastery, that does 3 dice of weapon damage, gets 2 Mastery dice, and can make up to 2 additional attacks if you kill targets, also fits right in at level 11.
As a side benefit, opportunity attacks don't suck (your mastery die applies to them, and you get extra damage dice), and two weapon fighting also doesn't suck (as it gets extra dice of damage as well) unlike baseline D&D.
We'd have to tweak the Paladin's Improved Divine Smite (I'd probably make it +2d8 radiant on a hit once/turn) to account for 1 attack/turn, and the same with the Barbarian (I'd swap Rage Damage for Rage Damage Die to start, then pack more oomph into their crit abilities).
---
Applying DPS reasoning to spellcasters, one thing I've played with is making powerful spells take more than 1 turn to cast. Like, instead of fireball being 8d6, it is 12d6 but requires 2 turns to cast, with a cantrip before starting casting and one after you finish casting it. (If you can cast a cantrip and start casting a spell on the same turn, spells like blade ward become interesting!)
The narrative impact of powerful spells gets spread out over multiple turns (keeping their damage per round under control), while the damage per second is kept up by making the spells impact larger. And wizards only spend 1 turn in 2 picking the ideal spell to cast.
---
Anyhow, I digress.