I wouldn't hold out much hope of getting a good combined
DC+clock noise, they would need two different analyses.
But I expect you will find that one dominates so much,
that one could be neglected. My bet's on the clocks.
One question is, is there any mode where the comparator
has a straight-through high gain continuous-time functionality?
If no, then you can't really get at LF noise.
When clocked it's likely that LF noise will be aliased up near
the clock frequency, so maybe LF self geneated noise is irrelevant.
For kickback I recommend setting it up as application-realistically
as possible; the degree of kickback voltage depends greatly on
the input terminal impedances, as does the settling back from it.