If I find myself in the situation where I cannot identify the shadow projected in a photo, should I assume them based on the theory of light and shadow and form them on my own?

This is correct. While being able to see the shadows in your photo can be useful, the important thing is to pick reference images where you can clearly understand the nature of the textural forms that are present. That is more important than visible cast shadows, because understanding where those forms are is going to be more useful in actually being able to draw the shadows in your studies.

As for the water droplets, that's way out of scope for this exercise, and this course. Since we don't get into actual matters of rendering, it's best you just stick to things with identifiable, opaque textural forms.