So tracking a star that passes behind the sun is seen to be above the sun and below the sun. Are these direct observations? Or is it because theory predicts such due to gravitational effect of the sun on light that these are direct observations?
That may be an indirect observation. It's tricky. I have to keep in mind that an observation that cannot be observed directly may nevertheless stand good in science as a direct observation.
Recall my mirror example. If I do not observe the person walking up behind me but nevertheless do directly observe the reflection, then the indirect observation (by dictionary standards) is a direct observation in science.
Just how far science considers the scope of an indirect observation, I'm not sure of. If science claims that something exists but has absolutely no direct observation at all and base its claims on the observation of other things, then that's a bit different than your star example because although we may not directly track each movement of the star, we have been tracking it. Using math and knowledge for saying where it's at while outside of immediate detection may (or may not--I'm not sure) be an indirect observation.
It's not like anyone is denying the existence of the star when its position is behind the sun, so our claims of its position when not being able to literally detect it may stand good as a direct observation because of our ability to mathematically calculate its position despite an ability to use instrumentation to detect it at the time.
The major point I've been trying to drive home all along, however, is that the answer is glossary dependent, not dictionary dependent. Whether or not something in science is a direct observation or not hinges on exactly what a direct observation is, and to know that, we should consult a glossary that defines the term. It is a mistake to consult the dictionary and consult the meaning of the terms, "direct" and "observation."
Since some indirect observations, lexically, are direct observations, stipulatively, it's going to be difficult for some to make sense of your claim that "all observations by humans are indirect". Suppose you're right and no human observation is direct. That is not the same as saying all observations are indirect observations (based on glossary standards) even if it is true that all observations are indirect observations (based on dictionary standards).