Heya,
I'm afraid this is a little bit toooo much of a simplification. While Mendels Rules can be partly applied to selfing, it isnt such an easy fit. Have a look here: https://academic.oup.com/jhered/article/107/7/635/2622950
There are also some aspects you cannot consider, since they are totally random. Crossing over happens all the time. Plasmatic heritage, heterosomal heritage, polysomie and some other things I cannot go into depth also.
The Dunnett square in your example is only working with two traits. There are thousands of genes and a lot of them are involved in the phenotypical expression.
You also have to differentiate between genotype and phenotype. Dunnett square shows you 16 genotypes, but 4 phenotypes. Which still doesnt mean that you will see all combinations by 4-16 seeds. Selfing a F1 can result in big surprises and isnt calculatable. Just pop a few and then a few more. Can be gold but also can be shti. You never know. But dont expect homogeneity. There could be 2 meter high sativa leaning plants as well as 50 cm one-headbud-indica plants. Im really excited for your results. Keep us updated
