I suspect there are factors in how cars are grouped we haven't thought of.
I could see some features wouldn't make a big difference. Some are the same hardware and things are just enabled by software like the 48 or 72 A charger and AP. I would think the colors interior and exterior would want to be grouped. We also don't know for sure if two cars started down the line on the same day. Tesla isn't good at updating customers, some people get updates on the site at every step and other people see things hang in one step for a while then leap forward. This may be due to some human updating status and some are better at it than others.
They have a lot of space in the plant, they could assemble a number of chassis in white, paint a batch of them a given color, then set some aside because they are going to get batched with some other feature set, and others go on the line right away. With a multi-day process, there are many variable ways of doing things that we can't figure out with the dribs and drabs of data we get.