If you listen to his whole rendition he adds the "ghost vowel" in several places, the "et" being the only indicated liquescent among them.

I suspect this is partly a French singing practice/habit. Here he applies it in ways that afaik French people usually wouldn't when singing in French ("super-ə"), but I'd suggest this practice is, among Francophones, just in the water.

I notice he also - and not only in this recording - merely gestures ornamentally at some neumes without clearly landing on them, and occasionally outright skips complexes according to some logic of his own. Not all that troublesome, but it doesn't scream "I'm paying close and scholarly attention to the notation."

This style actually strives to not fixate on the notation, so I'd further suggest that, whether or not anyone wants to copy or be inspired by it, it's the wrong place to look for "strict" and "accurate" interpretations of neumes, because he simply doesn't care much about all that, at least not in the more usual way. It's beautiful music though.

On the larger question, with a liquescent on "et" I wouldn't do anything in particular. With liquescents on voiced consonants, in contrast, I tend to give fairly generous space, strength and expression to the consonant. I'm pro-consonant in general: they launch and land the vowel sounds and have a lot of character in their own right. I don't think of them as second-class citizens of the sound world, and I don't personally go in for really light and delicate treatment of them especially in chant.

This might deserve its own thread, but I do not think that the approach by the "École Grégorien" can be nonchalantly dismissed. Obviously it is heavily inspired by Ensemble organum (Marcel Pérès), which is based an a lot of research. Apart from the much more solemn tempi, the most obvious difference is in the realm of oral tradition that was never written down: portamento and ornamentation. Portamento (gliding into a tone from below) was so ubiquituous until the first half of the twentieth century that it was not even mentioned back then. It is only noticable retrospectively, because it makes old recordings so much different from modern recordings. Somehow it came out of fashion and is today considered bad teste (I must admit, though, that I like it when applied as an ornament to stress certain notes and I feel that we are today deprived of an important means of musical expression): Even Marcel Pérès stops short from it.

In his chant method (1919) Dom Lucien David, an advocate of the Pothier approach (Old Solesmes) favors the shadow vowel (option 3). I think this practice could help with the projection of the words in a very resonant acoustic.

"Stop-stop" liquescences (t–d, b–d, etc.) and portamentos are neither confined to French nor to Gregorian chant. They are used also in folk songs for articulation or expressive purposes, also in some forms of vernacular psalmody-like chant. That such liquescences find their way into the chant manuscripts may just mean that they were sung that way. In some late-medieval manuscripts there are even liquescences on consonants followed by a vowel, e.g. bonum est in the offertory of Septuagesima. Of course, copyist's mistakes cannot be excluded.