The word "sŭŭs" is always counted as a sequence of two distinct vowels in latin hexameter, as you can see, for example, in Verg. georg. 4,190:
In noctem, fessosque sopōr sŭŭs ōccŭpăt artus
in Ov. ars 2,643:
Nēc sŭŭs Andromedae color est obiectus ab illo
and in Ov. met. 2,186, which has sŭūs just like your verse:
Frēnă sŭūs rector, quam dis ...
Here's an example from Lucan's Bellum civile (8.321) where īt is used and ĭĭt would break the meter:
nomen abit aut unde redi maiore triumpho? (8.321)
The form abiit would produce three short syllabus in a row.
There's not much literature on this subject -- maybe because synizesis is so rare. Aside from the obvious answer of "whenever the meter requires it", I couldn't find any hard-and-fast rules, just hints.
However, from what I could find, it's particularly common in...
the initial ea- and -eo of eadem and eodem.
words ending in -ea and -eo
the eu, ei and ea ...
Synizesis of ee is supposed to occur in forms of the verb deesse. Presumably the result was [eː], with the same pronunciation as ē. This seems very similar to the contraction seen in words like dēbeo or dēmo.
Evidence from poetry indicates that those imperfective forms of deesse ‘be missing, absent’ where the stem begins with [e] are contracted even if ...
I went through the first two books of Aeneis and found no evidence of this kind of synizesis.
These two books contain 25+31 perfect forms of the kind (with the ending starting with a short 'i').
Some of the 'ui's could be parsed as either two short syllables or one long syllable, but in most cases the metric forces two syllables.
There was not a single ...
I haven't read any linguist's comments on this matter, so my post is just a collection of guesses.
Perhaps for grammatical reasons
Elision frequently deletes vowels that are part of grammatical endings, or sometimes part of highly grammaticalized monosyllabic words such as est (with "prodelision"). Grammatical endings often have a certain amount of ...