1- Shifting out of form to throw a heal or two arround once per fight isn't going to have a significant impact on raid DPS.
2- Mana is a non-issue as balance. Your moonkin probably specced out of every mana regeneration talents, except for maybe 1-2 points in Intensity in addition to Omen of Clarity.
3- Are we talking about the occasional rejuvenation followed by a nourish, or is he staying long periods outside of forms?

As for an "hard and fast rule", staying alive is usually the best way to make your DPS go up. When I get into trouble in a raid damage heavy fight (Think HM Mimiron, HM General Vezax, HM Twins Val'kyr), I usually Barkskin->Healthstone, then proceed to heal myself with a quick nourish (or get a tick or two of tranquility if anyone in my group is also in trouble).

Of course you can tell him to trust his healers, make him play like he was a pure dps class without the ability to heal, yet they still survive most fight. But pure dps classes (and most hybrids) have very good defensive cooldowns (Iceblock, Cloak of Shadow, Bubble) while we're stuck with Barkskin (and possibly healthstone) as our only survival cooldown.

Just feels like there is way too much reliance/vested dps build into Eclipse. Its cool to change rotation, however if our competitive dps was build upon one talent (and set bonus) then that isn't a great design at all. That dps should be shared into other suitable talents.

I guess we'll understand more clearly once the nerfs hit; but going backwards in dps doesn't help future invitation to raids to get the gear to "claw our way back to the DPS we're currently doing".

It's odd to think that blizzard would remove the benefits of WE just to force us to play a time-waiting game with the last starfire. I believe Blizzard is trying to enforce the idea that you should be proc'ing each eclipse by virtue of the spell's natural crit (not by exploiting the more detailed mechanics of the eclipse mechanism). Trying to time your starfire to take advantage of client-side buffs/internet-server lag is similar to cheating, like wave dashing in smash bros. It's an odd, but very favorable, feature that was surely unintended.

But yeah, I don't think the "fix" that blizzard implemented is a very good one as well. As people WILL attempt to take advantage of it anyhow, except now it's just harder to take advantage of. The problem still exists, just in a smaller (yet still reasonably exploitable) form.

Along with the other major hits to the moonkin, I just wish they would provide a much better buff to moonkins other than a 2% crit buff on the final 4-tier bonus. I don't think we are OP, and we deserve a nice buff to the 2pt10 like longer proc durations, having durations "stack" on each other instead of overriding, or both!

I did some testing on my moonkin the other day with proczilla and I kept getting about 1.3 clearcasting procs per minute. This is much lower than the 3.5 ppm (or even 2 ppm) that I've seen posted in forums. I was testing for about an hour. Was I really unlucky? Is proczilla not working appropriately? If there had been some previous testing done, is there a link to this test data/post?

Couldn't find this on any page, but i will keep it short:

Regarding 4pt10: Ignite can crit, right? Can languish?

Ignite cannot crit.

I have a slightly "hybrid" take on Dachus' point, which is - if I get correctly - should (and when) a Moonkin heal at all?

One of the benefits of having roles defined in a raid is "expectability" in the gameplay. In other words, if you heal, you know you're going to cast a bunch of hots before your mates start taking damage, etc. This is obviously key to the success of an encounter, except maybe on weirder fights like Faction Champions, but that's another story.

My point here, is that although having good reflexes, and occasionally saving one of your buddys' ass is great, healing the main tank is not generally your job: in short, do it if you are supposed to do it at some point during the fight, don't do it out of "omg tank has low hp." Because if you do, your raid loses a bit of that expectability, and it may affect progress.

Healers need to learn when the tank is at risk, when the raid is at risk, etc. If you "reflex-heal" your mates, you affect this learning process in an arguably negative way.

However ! If there is an encounter that can be made easier with a plan that involves you healing, then by all means, this is a great thing to do. Operative word here is "plan". I'd also like to hint at the fact that tranquility, because it is such a unique spell, is probably what you want to plan for in 90% of the cases. I remember, as an example, than Sartharion3D (10 man) speedkill required the moonkin to pop tranquility at some point during the fight to ensure the raid survived. From the top of my mind it was when d2 and d3 were active, or somewhere around that time (I was not moonkin at the time).

Interesting question, hope my answer will be useful to it.

Some interesting words from GC on the T8 nerf and Eclipse in general:
MMO-Champion BlueTracker - Rogues get a buff. Moonkins get a nerf?

To get to the point:
He does not go into details about how they are going to "Offset", but I suggest waiting with calculating upgrades and what our DPS loss will be until he does.

On another note, I don't think I've seen this mentioned here:
Edit:
Just posted with the guy above me, sorry for the seemingly double post

Well, for what it's worth, I made a post on the official forums:
World of Warcraft - English (NA) Forums -> WiseEclipse comments, from its originator

(First time I've posted there in like 3 years :/ ). Let's see if anything happens.

I haven´t thought it through thorougly, but in my opinion one of the biggest scaling problems for us moonkins is not exclusively eclipse, but wrath. Wrath goes below our gcd since months and we have to cope with the 400 haste-rating-softcap since the beginning of Ulduar. This could be easily fixed via glyphs, e.g.
Glyph of Wrath: Increases damage of your wrath spells by 20%, but also increases its cast time by 30%.
Actual percentage-numbers could be theory-crafted. This glyph would replace the IS-Glyph, bringing back the 3% miss-chance while keeping NGs functionality. The loss of IS´s damage might be counterbalanced in the balance-tree, e.g. a 15% dmg-increase, if needed.
The benefit would be a more "balanced" and scalable spell-combination of SF and W while keeping dots intact, plus giving the player a choice to glyph it or not, depending on his equip.

What do you think?

I was giving some thought about the Wrath issue last night. I don't really think increasing the cast time of Wrath would help, it would simply move the goal posts. What would be interesting is if haste were to effect Starfire and Wrath differently, i.e.

Starfire : behaviour as-is
Wrath : haste increases the base damage and mana cost (maintain same mana to damage ratio) but does not affect casting speed. casting speed increases still increase the speed of casting (i.e. bloodlust/heroism etc).

This would change the dynamic of the two spells as well as effectively removing the haste soft cap to a much higher number. It then becomes a choice of how much mana you want to burn in a given time as the more haste you have, the more wrath will burn.