Since it's a chance proc, you might not even get it during the first moonfire so that's another 2 gcds to reapply it, which you would need to do to make the spellpower proc worth it over the course of a 5 minute fight. And assuming a 45 second internal cooldown (correct me if I'm wrong), you usually wouldn't be able to time when you wanted it.

Let's assume fight duration in seconds (F)= 300 (5 minutes)
Spellpower of Heroic Phylactery Proc(S)= 1207
Duration (D)= 20
Average spellpower= X
Therefore X= 1207/(F/D)
A 20 second proc is 1/15 of a 5 minute fight, meaning that the total spellpower of the proc is averaged over the course of it. If your phylactery procced once over the course of the fight, it is worth 80.466~ spellpower and 172 crit, if twice then 160.933~ spellpower, making the overall value to a tree relatively inconsequential in comparison to other trinkets we could get from just about anywhere else. Even a Heroic Solace of the fallen is worth 158 constant spellpower and 90 mp5.

So while the idea is clever, it doesn't hold when the cards on are on the table and push comes to shove.

The real power of phylactery is an effect targeted at a certain point in time, not averaged over time. It's probably not the most reliable thing around, but it also puts a spotlight on the current trinket list, which includes no +sp clickies either (magitha's for ~700 sp for example). Damage in a fight isn't always aura based, in fact there are times where damage is very concentrated on one point. One example would be saurfang heroic as blood power rises. If mana regeneration is not an issue, clickies are still an option.

If cast 15 secs before a high-damage period it has a 75.99% (not sure if that's a rounding error or what wow would do with the number) chance to proc within that time (If i'm not mistaken, the procrate for phyl is 30%, thus bringing us to 1-0.7^n decimal procchance) in the "worst case scenario" aka a proc at 3 seconds after cast, with 12 secs remaining to that period you will have 8 seconds with +514 (usual +healing talents, may vary somewhat with spec) on each rejuv tick. It's not quite dependable, but a one-in-four chance of failing is still an interesting prospect. regarding not having free gcd's, one has to consider the worth of a WG when a lot of heals are being directed at the melee already / the raid is spread. If the damage at that moment is lower, then spending the gcd might not be a major issue.

With that said, I'd agree to lending it to the dps over using it yourself. Just a bit of a thought experiment here.

I am unclear as to where you are getting your numbers from. Phylactery has a chance to proc when "one of your spells deals damage", therefore there are 5 chances for it to proc off of one Moonfire, the initial damage and the 4 subsequent DOT ticks. The chance is not cumulative and phylactery cannot proc without an offensive spell first dealing damage. Therefore Phylactery has five 30% chances to proc, not a 75.99% chance to proc over the course of those 5 times. Statistically, yes that would make sense, however that is not how chance mechanics are designed. It is not a one-in-four chance of failure, it is a one-in-five chance of success.

Regardless of what damage is going out, casting moonfire essentially costs 2 GCDs: One to cast the spell and one to shift back into tree form.

It has a chance to proc on spell periodic damage. I'm not completely familiar with moonfire mechanics, but I believe the first hit is direct damage and doesn't count as a dot tick. I don't quite understand what you're getting at with the chance mechanics here - first you say that it has a one-in-five chance of success (somehow based on the amount of ticks in a moonfire), then you say that each tick has a 30% chance to proc. What I mean is this: while the length of spellpower bonus remaining will vary after "high damage period" begins, you will have a 75.99% chance of having at least 8 seconds of spellpower bonus left after it begins ("worst successful case"). I'm not getting my numbers from testing, but since it seems to be an accepted fact that it has a procrate of 30% on a periodic tick I use that for my calculations. If the procrate is only 30% per spell regardless of the amount of dot ticks, I was unaware of this.

You may be thinking in terms of gambler's fallacy when you should be regarding it as a set. It's not a choice of whether to "flip the coin again" here, you will always have those four ticks, which will give the trinket the calculated probability of proccing.

I missed the two GCDs point though, I'll admit it makes this seem even more silly of an idea. Probably best to just leave it alone for now.