This is an interesting question.
I can not tell much about the actual usage of this expression, but I will make an attempt to explain why it could be used that way. My answer is probably more a philosophical than a linguistic one.
If you say you are familiar with logic and interested in philosophy, you might have made acquaintance with modal logic:
◻p means "it is necessary that p"
◇p means "it is possible that p"
Without going into too much detail about the logical spell-out, one way to formalize these modal operators is the assumption of so-called possible worlds, which are our metaphysical space:
One can imagine a network of possible worlds, where each world is a very, very large situation, fully specified to every detail you could make an assertion about - there is our world of course, then there is one world which is almost identical to ours except for the fact that that bike standing out there is not red but black, then there is one world which would have been identical to ours except for the fact that Hitler never existed, leading to a completely different development of historical events, then there is one world which is completely different to ours w.r.t. the fact that not humans but giant ants have taken over world domination, ... - and as you can imagine, there are in theory infinetely many of such worlds, already due to the fact that our bike could stand one meter from the wall, 1.20 meters, 1.232323 meters, 1.23932545565463 meters, ...
Our universe is a network of many of such worlds, out of which some can see each other, others can't. For example, if our universe consisted of worlds w1 to w10, we could have the case that w1 sees itself, w2 and w4, w2 also sees w1, w3 and w10 but not itself, w3 sees w1 and w2 (so we have a reachability circle of w1, w2 and w3), but nothing else, w4 sees w8 and w10, and so on.
Note that this model does not make any claim about the actual existence of such parallel universes. It is really only there to capture the notions of possibility and necessity in somewhat more graspable terms.
In this network now, "necessary p" means that from our standpoint (i.e. the "actual world"), in every world we can see, p is true. It might be that there are worlds in which it is not true, but these worlds are not reachable to us, so in our world, p is a necessity. For example "It is necessary that when you drop a pen, it falls down" - it might be that there is a world in which this is not the case, but at least this is in reach of our imagination.
"possible p", on the other hand, means that there is at least one world of which p is true. For example "It is possible that this winter will be a cold one". In our network of worlds, we can imaginally reach a world in which the approaching winter is cold, so it is possible that p is true, but we might also be able to imagine worlds in which this is not so, thus p is not a necessity.
Note that these "alternative worlds" can also include the actual world we live in, if our world is defined as being able to itself in the universe. It could be that p is true in none of our neighbouring worlds, but in our own one, and if the network states that our world can see itself, then p is possible in our world.
Formally, where R is the reachability relation between worlds,
⟦◻p⟧w = 1 iff ∀w' : if wRw' then ⟦p⟧w' = 1
("It is necessary that p" is true of a world w if and only if in all worlds w' which are reachable from w, p is true)
⟦◇p⟧w = 1 iff ∃w' : wRw' and ⟦p⟧w' = 1
("It is possible that p" is true of a world w if and only if there is at least one world w' which is reachable from w in which p is true)
Now what the "philosopher's English" expresses is exactly this: They can imagine a world in which the speaker's hair is green.
The reason why this might sound odd to the ordinary English speaker's ear is that we don't usuaually talk about such metaphysical possibilites, but about the facts in our actual world.
The necessity and the and the possibility operator then relate to alternative worlds to our believe, i.e. our epistemic state. In the world which I am currently believing to be true, my hair is not green, but it could be that I am mistaken and the world in which I am living actually turns out to be one in which my hair is green. In my head, these are different states of affairs, but it still all is within the scope of our real world, and the evaluation of necessity and possibility only relate to the correctness of my believes.
The philosophers, on the other hand, don't speak of epistemic, but rather about metaphysical states; the epistemic facts all hold true in our actual world, and the alternatives take place in worlds which are actually different from the one we are currently living in.
Now since the "normal" past is, in ordinary English, already occupied by the use for epistemic states, a different tense needs to take over the role of expressing metaphysical possibilities.
This is where the plupferfect comes into play: In normally expressing something even further in the past, it adds some distance to what is being said; instead of expressing that it
could be the case (as a matter of fact in our actual world, if your epistemic state does not match the actual circumstances), you now express something even more remote, namely that it could have been the case
if the world we live in wasn't ours, i.e., if we were situated in a metaphysically different state.
To make it easier for you to grasp the idea, just imagine the pluperfect proposition to be the consequent of an if-statement, like
If we had been born in a different world, my hair could have been green.
To summarize:
My hair is green
-> This is a fact in the actual world.
My hair could be green
-> This is a reachable possibility within my epistemic states, which still all revolve around or actual world.
My hair could have been green.
-> This is a reachable possibility on the metahpysical level, if the world we live in wasn't necessarily our current one.
Thus, could word this grammatical phenomenon as "when you switch tense, you switch levels":
When you switch from present tense to simple past, you switch from one-world level to the level of different epistemic states in our actual world.
When you then switch even further from present tense to pluperfect, you switch from little epistemic worlds within our actual world to the metaphysic level of different worlds out of which our actual state of affairs is only one world out of many possible worlds.
In that respect, I find the work distribution between English simple past and pluperfect quite clever, the temporally closer one expressing a possibility in terms of - mentally closer - epistemic states, the temporally more distant one expressing a possibility in terms of - mentally more remote - metaphysical states.
This is really just a hypothesis, but the explanation works for me.
