topics_over_time top terms less than top_n

Hi,

I've noticed that when I call `topics_over_time` it generates topics with less than `top_n` representative terms. Initially I thought that this could be because very few documents were assigned to these topics, but I see that it happens with the topics with many documents in the time period as well.

Is there any way to know what causes this?

Start by reproducing the reported topics_over_time behavior and inspect the generated representative terms for topics with different document counts. Determine what causes fewer than top_n terms, then document the cause or identify the required behavior and verify it with a focused regression check.

Written by the indexing model from the issue text.