I was thinking again (yeah bad thing to do) and it came to me that realistically a non-dot class is always going to be burst based eventually (it might take some setup, and the setup period can be used to control PvP burst, however eventually it will need to be capable of doing N,000 DPS).

Thinking this, with Retribution hoping to sit around 6,000 DPS at T8.5 made me think of our rotation:
9 6 9 6 9 6 ... repeat, roughly 9s rotation (slightly longer to get the full pattern in place since the abilities swap position but basically a 9s rotation). Consecration handily lasts roughly this long, and is damage over time based.

So 9s duration, 6,000 DPS ~ 54,000 DPS (2 CS, 3 3.0s melees, 1 DS, 1 Judgement + seal procs) and I started to see an attack pattern more like:

Consecration---------------- 18,000 Damage over 9s ~ 2K DPS
|.Divine Storm------------- 4,500 Damage over 9s ~500 DPS
|.|.Judgement------------- 3,000 Damage, 6,000 Damage DoT (9,000 over 9s) ~ 1K DPS ~ 3,500 DPS at this point with low burst
|.|.|Crusader Strike + Seal Proc + applies debuff
|.|.|Crusader Strike + Seal Proc + applies debuff
|.|.|Crusader Strike + Seal Proc + applies debuff

The remaining 2,500 DPS to be made up over 3 GCDs (2,500 *9 = 22,500 damage per ability, roughly 3,500 CS, 2,000 Seal).

The rotation can be made interactive via procs (Art of War, Vengence, whatever) subbing in spike damage or large dots in place of the CS portions. At this stage the retribution Paladin cannot burst effectively as there are multiple options to forward / backwards setup buffs:

Consecration is the highest priority, it buffs all other damage. Divine storm then sets up a 3-5 stack of debuffs that can be taken off a target (setting up a CS string the same way death runes do for Blood giving the HS HS HS HS [HS HS] string if desired). Finally Judgement sets up CS to actually activate seal procs making Crusader strike the most damaging part of the rotation (however requiring the setup 3GCDs to actually perform well). If backwards synergy is desired, CS can setup a buff acting to improve damage done by non-CS sources meaning your rotation is always best served by mixing CS and other strikes (so the CS CS CS CS CS CS plan is a bad one, as is Judge CS CS CS CS CS CS).

Procs such as art of war feed into making the spec bursty, proccing nice damage boosts or tricks like exorcism.

This further compromises PvP burst by harming mobility, without an active consecrate under them the Ret paladin loses damage and thus becomes less capable of burst by default. This functions like desecration however without the snare making Ret's kill the use of HoJ to keep a target within Consecration (ticking at ~ 2K/tick) and enabling the burst setup. This also means casters must move from ret (standing and pounding is dangerous), and forces gap openings on melee combats to counter ret's lack of MS or similar (standing near Ret will hurt even if you are hurting them, ergo encouraging attack and feints).

On PvE fights, the short duration of the Rotation (9s) means a Ret forced to move is not overly penalised (they will get into combat, drop Consc within 9s and get upto speed in the rotation), and you are not penalised for swapping target overly other than the loss of the reverse synergy between the CS debuffs and DS/Consc/Judgment. Retribution at this stage has massive AoE DPS potential (2,000 DPS from consecration per target, in a situation like Freya with 15 targets would see spikes of 30,000 DPS) requiring the addition of a damage cap similar to that of other AoEs or perhaps scaling with the Paladin's stats (so a 2,000 DPS consecration caps at 15,000 DPS + crits, while a 2,500 DPS consecration caps at 17,500 + crits) to stop the problem other caps have of making the spell stop gaining power as you gear up.

A 9s rotation is quite short, however it is reasonably comparable to that of a destro lock:

|.Chaos Bolt
|. Filler
Repeat to keep immolate up,CB / Conflag on cd.

Ours would require stricter control and the correct substitution of procs to make it effective rather than the Destro lock badly overlapping CD model.

I would argue that this model is already applied via Affliction locks and similar having an effective rotation (ok the dot timers line up horribly) similar to:

Corruption (should only need cast once, the exact order of this rotation is wrong but the idea is there)
Unstable Affliction
Curse of Agony
--- Filler
Unstable Affliction
--- Filler
Curse of Agony

The DPS of the spec is instantaneous (it is in its full swing from the second Haunt is up) and maintains this easily. The filler portions are due to the badly overlapping timers of the dots (which could be closed up and made into a perfect rotation like the 969 or my proposed Ret).

Ret at this point would have lowered burst, and high DPS potential, however still lacking is a reason to break the rotation in PvP. The obvious solution to this is to alter the functionality of DS and Judgement making one of these an interrupt (I would suggest judgement for this setup since in my vision it enables the seal procs on CS), and the other into a mutating ability such as the use of Death Runes such that it can be worthwhile losing the setup damage from the DS in return for the mutated ability proccing from it in certain situations. In this case something like:

Divine Storm: Deals 110% weapon DPS*N to M targets (including current target) and heals for (M, max 4) * damage dealt * K. The paladin gains "awesome buff" for 9s increasing Holy damage by L%.

Paladin MS: Prevents all healing to the target for 4.5s (GCD + 2 Strikes), requires the Paladin to not have awesome buff.

In effect a forebearance effect: PMS > DS works, DS > PMS does not, meaning if you expect the target to be healed you break your damage rotation (substituting in CS with no seal proc, or an early judgement) for the ability to interrupt the healing done to the target.

This finally puts us in the right position:

Paladins have a simple rotation
Paladins have a modifiable rotation (proc based so its not a simple macro dps)
Paladins have a reason to reduce their DPS potential for utility gain
Paladins have the ability to kill healers / other targets by chaining Stun / MS effects, however need high time on target to do so
Paladins retain the utility at the cost of damage model they have currently.

One major issue which crops up with this solution is the power of consecration. Dealing 18,000 damage over the 9s is ideal to reduce single target burst damage while maintaining overall damage (if you need any help quelling doubts go do the Illidari council fight, his ticks for about 3,000 iirc and is very effective at denying an area but does damage slowly enough that you can escape with only 1-2 ticks at most).

The issue with this is a fight like Kologarn (and to a lesser extent Freya with the adds, however in that case a simple DPS cap on the ability will come into play due to the number of adds), where there are N large high health targets such that the consecration damage is multiplied up in a useful way (Kologarn + active arm = 2K bonus free damage, + health on arm to blow off for last transition means the health taken off it is not useless). This implies some kind of diminishing returns on targets / a target cap such that:

1 Target: 2,000 DPS
2 Target: 3,000 DPS
3 Target, 3,750 DPS
Damage cap applies and reduces effectiveness.

If this is seen as overly powerful, the buff effects can be affected by forbearance making bubbling a real choice rather than a DPS drop for the duration of the immunity. This means that you can effectively tag the paladin into a low DPS support mode by forcing the bubble, however doing this is unlikely given the availability of other similar cooldowns.