Is Proto-Balto-Slavic zero-grade from long zero-grade i
Is Proto-Germanic zero-grade from long zero-grade u
As three seconds of looking it up (probably in the same place you found those forms to begin with) shows:
- PBS *pílˀnas < PIE *pl̥h₁nós > PGmc *fullaz
- PBS *wilkás < PIE *wĺ̥kʷos > PGmc *wulfaz
- PBS *źírˀna < PIE *ǵr̥h₂nóm > PGmc *kurną
- PBS *śírˀnāˀ < PIE *ḱr̥h₂-néh₂, *ḱr̥h₂-nó-m > PGmc *hurną
So evidently the *i and *u are epenthetic and the result of the devocalisation of the vocalic resonants. That is to say, Proto-Indo-European *l̥ and *r̥ represent /l/ and /r/ functioning as vowels, which is something neither Proto-Germanic nor Proto-Balto-Slavic likes, so they convert them to consonants and add a vowel to avoid the loss of the syllable and further pronunciation difficulties.
"Zero-grade i" and "zero-grade u" aren't meaningful terminology. Your table shows the behaviour of an ablauting vowel and a following semivowel in Proto-Balto-Slavic; when the ablauting vowel is present, those semivowels exist as glides, and when it's not (i.e. when it's in the zero-grade), they turn into vowels. In this case the roots of these words are in the zero-grade, but they don't have following semivowels, so it's not relevant.
(There are other languages that deal with the *l̥ in *wĺ̥kʷos by vocalising the *w instead: Proto-Celtic *ulkos and Albanian ulk, for instance.)