I have to agree here. There's no reason that damage alone has to be the only direct contributor to encounter completion. Take the green dragon fight from ICC in WotLK, for instance: the encounter completes when you (nearly) top her off. This would simply be content with a different victor condition. Alternatively, if a Very Important Mob/Person's survival over a given time limit, or over a given escort distance, were a victory condition, it's likely that gimmicks may be necessary to make that encounter satisfying in its interactions, such as by creating more ways to provide mitigation for the mob (beyond current interception mechanics) where topping it off and curative-shielding it would otherwise be insufficient to prevent its being one-shot, but the victory conditions themselves still wouldn't be a gimmick, just an alternative to the most basic characteristic of the encounter.
[1] Shelltron was one of the mechanics that upon implementation had me begging for more of similar type -- albeit with slight revisions. More on this later, in all likelihood, as I feel they are relevant at least to increasing the usefulness of tank toolkits, which otherwise suffer, in a sense, reverse gear-scaling (relative, at least, to healing). And I definitely agree that these point out interesting directions that are yet barely, if at all, touched upon.
[2] If I recall correctly, from my HW days of tanking just way too many dungeons daily, there are actually many enemies that do follow strict rhythms to their specials, be it down to the second within uptime, down to the second regardless of uptime (which removes some advantage from kiting), or after a set number of auto-attacks. Additionally, some either strike immediately after a given auto-attack (or potentially simultaneously), or after a brief delay. I actually tried some old dungeons like Fractal solo on DRK to see if I could break TBN in Grit with my full gear's worth of HP, by slightly moving mobs about so that their next specials would overlap predictably. So, in a way, this is already kind of in XIV, but it remains painfully hidden, to the point that you could say that for the majority of players, these concepts are only latent in the game, rather than a real part of it.
The concern that pops into mind first though, is that this game is really rather particle-effect and animation-heavy, almost to the point of obscuring natural telegraphs — requiring our slightly gaudy hit-area indicators. When the majority of players are under the assumption that there's nothing they can do to better time their tools against these specials, they don't feel obliged to turn off their particle effects and dart their eyes about to see them, which to many — not all, of course — will be less enjoyable an experience. Similarly, audio cues (while solvable via visual sonar for deaf players) would be insufficient to differentiate between special wind-ups between two adjacent mobs of the same type. So as much as I want to make the sound of and vision over our enemies actual sources of information, there's currently a lot standing in the way. Maybe we'd need some sort of... post-processing (?) of our particle effects to allow for conditional reversed layering, as to allow enemies winding up or giving other visual tells to be visible to the player's camera through the particle effects. Maybe some more generalized sound-to-vision cues without needing visual sonar, specifically... I don't know. I do reeaaally want this though.
[3] This would be such a huge boon, imo, to the tanking experience. I'm not sure if pure manual manipulation is the best way to go about it, but I feel there definitely should be tools to allow for this. More on this later.
[4] I'm somehow who generally hates QTEs, and I still think some form of that could be awesome. Heck, I wouldn't even mind seeing it as a general mechanic of sorts, as long as its soft enough not to feel wholly tank-specific (just... "tanks do it best").
As for the netcode... I've no idea either. They seem to leave more than most games do server-side, yet they still have bots capable of teleporting about underground, so...?
For a short time, there was a "Guard" ability on a very short cooldown, during which you could not attack, but could guarantee blocks for some total absorption amount. At that time though, it wasn't actually even part of Gladiator, but rather part of a side-class called Sentinel, which was usable by anyone who could use a shield, including casters. It was removed when Yoshida took over, leaving only Aegis Boon for burst-mitigation (converts a single strike's damage taken to healing, like an up-to-200% mitigation ability, on a 30s or 45s CD, I forget which).
Something like this wouldn't have any impact on or by the game's having or not having a GCD... which the netcode takes surprisingly little advantage from, considering the length of its animation locks. (We have a minimum true global cooldown, or "animation lock" of .5 seconds, generally around .7+ for oGCDs or .85+ for weaponskills, iirc, which should be enough in which to queue any ability without clipping even on a 240 ms ping connection, and yet at that same ping will delay a Suiton by as much as 2 additional seconds at a 2.35 GCD, when it should only delay it by ~.7.) Unless these were introduced as a weaponskill like 1.0 - 1.15's Guard skill — back when rate was determined more by Stamina and cast times, alone, than GCDs anyways — these could remain almost wholly fluid by simply being oGCDs as per Shelltron or TBN, or so I've read it. (Please correct me if I've misinterpreted, Singularity.) Even if made weaponskills, or psuedo-GCDs like the HW Empyreal Arrow, however (therefore coming at the more general cost of time, rather than specific resource or MP) their gameplay shouldn't feel in any way out of place or unsatisfactory in a GCD system. It just might mean that you would have to waste a bit of time to line up the active block occasionally, which wouldn't feel great in itself, but might even go so far as to help you feel like you're attacking something that isn't just a script dummy for your own memorized rotations.