and also please guide me how to acquire such information from online linguistics resources (for example WALS.info).
Modern Persian is known for this, hence
Isfahan from earlier
Spāhān, and I assume some other related Iranian languages share this phonotactic constraint.
Western Romance languages have tendencies against initial
sk-, similar to Persian (and Turkish). Thus they have typically inserted
e- (and many even drop the
-s- in writing or speech, see for example the descendants of Latin
schola), Ladino, Walloon and Sardinian go both ways. Some Eastern Romance dialects changed or change
ʃ- before some consonants, similar to High German.
We should make a distinction between tendencies and real constraints, for example modern borrowings into French do preserve the clusters in words like
sphère and so on. Likewise the English rules for loans are a bit looser, whereas Spanish is stricter, but like English it breaks with the written form even in prescriptive use.
Every language has some tendencies ie frequencies and it is hard to know all regional spoken variations, but Slavic, Baltic, Germanic, Armenian, Celtic, Albanian, Hellenic are mostly happy with consonant clusters (although Greek has no
ʃ and thus no
t͡ʃ, German has no
ʒ and thus no
d͡ʒ and so on, and of course German has
ʃt- but no
st- in native words).
So my sense is that simple binary categorisation is not possible, and the remaining candidates are other Indo-Iranian languages, or non-Romance creoles or mixed languages with strong influence from Western Romance.
One method to check initial patterns is to seek in the dictionary, for example en.wiktionary.org/wiki/Index:Scottish_Gaelic/s#st, being sure to check the IPA pronunciation as languages like English and Spanish do not necessarily pronounce these even when written.