What is the likelihood that long-lost books, such as Anaxagoras' book, Aristotle's 2nd Poetics, or Homer's Margites will ever be found?

Pretty close to nil, I fear.

Here's the thing: long-lost works do turn up now and then. It's rare to get intact ones, but it does happen. Two years ago a bunch of Origen's commentaries on the Psalms turned up in a manuscript that was already known, but the author hadn't previously been identified: that's probably the biggest one for a few decades. Then going backwards in time there's the Philodemos papyri, the Derveni papyrus, the Dead Sea scrolls, Menander, Bakchylides, the Vatican epitome of pseudo-Apollodoros, and others that I'm not thinking of. So these things do happen. And fragmentary ones pop up still more often (new fragments of Menander and Hesiodic poems pop up all the time; new Sappho fragments in 2014 and 2005; a new Archilochos fragment in 2005; Simonides fragments in 1992).

But, and this is a very very big but, that's not the same thing as finding a copy of a long-lost book that you happen to be interested in. Long-lost texts may turn up now and then, but no one has any control over which long-lost texts they're going to be. And the pool of potential long-lost texts is really really big.

So even if a complete copy of a lost philosophical work turns up, say, at Herculaneum (once someone works out how to image where the ink is inside the scrolls without tearing them into tiny shreds), it's very very unlikely that it's going to be Poetics book 2. It's enormously more likely to be a piece by Rabirius, or Philonides, or Diogenes of Tarsus, or Basilides. And while that'll be nice, it won't be quite what you're hoping for.

So yes, it's very possible that a new Greco-Roman philosophical work will turn up one day. But hoping for a specific work by Aristotle is a very, very long shot.

The other examples you give are also long shots, I fear. Anaxagoras and the Margites are just too early to be likely to have survived in any intact manuscript anywhere. Very nearly all intact manuscripts come from the Roman era or later, and in the Roman era there just wasn't much antiquarian interest in preserving copies of things just because they were old. Only a handful of older works got into "the canon", like the Iliad and Odyssey. The things that survived, survived because they were being actively used for one purpose or another. Plato and Aristotle have an OK chance of surviving in that context; Anaxagoras, unfortunately, doesn't. And only a few pieces of pre-Hellenistic poetry (other than tragedy) made it into "the canon". (We should consider ourselves outrageously lucky to have intact poems by Pindar and Bakchylides, because that's exactly the kind of stuff that tended not to be preserved.) Now, it's likely that copies of them were still floating around; just not many. So we've got an OK chance of finding scraps of them in Egyptian papyri. But only scraps.