This can be done in PALM and multi-level exchangeability blocks. Let's recapitulate: you have 2 groups, in which subjects have one or two visits. With this you can make group differences of two kinds: whether the two groups on average differ (considering on average both timepoints), and whether the two groups differ in their differences between timepoints (i.e., group by time interaction). You can also test for within-subject effects (time differences) within group, and on average across both groups.

It's necessary two designs for this: one for between-subject effects, another for within-subject effects. The within-subject effects require subjects that have both timepoints. Subjects who have just one timepoint provide no information about change over time. Please see a mockup of these designs here: https://s3-us-west-2.amazonaws.com/andersonwinkler/mailinglist/design_alex.ods. The 2nd design has the subjects who have just one visit removed (I retained their "id" for clarity).

Hope this helps!

All the best,

Anderson