Dual Branch Recurrent Neural Network, what is the correct architecture and can I turn off one branch?

Let's suppose I want to predict the daily consumption of apples in the next 30 days of a person and I have, as input, the historical information about the past 60 days and the demographic information of the given user.

I was planning of building a Hybrid NN with two branches, one dense and one recurrent, let's say made with LSTMs, that are then concatenated to a final branch. I was wondering, since this branch will have as output a time series, is it correct to make it dense? if not, how can I use a recurrent one if the input is not a time series?

Let's say, then, that sometimes I have to predict the next 30 days consumption of apples also for users that do not have historical information (maybe people that never ate an apple before) is it possible to turn off one branch so to use only the dense part?