From a purely methodological point of view, such retrospective diagnoses are to be abhorred unless there is more, although always circumstantial evidence.

Just last night I watched a documentary on National Geographic on Hitler's family (featuring Cesarani and Ryback as commentator's - remarkable heavy weights for such a documentary). The documentary made note of there being evidence of mental illness in Hitler's (extended) family. At that point, I became very curious to see whether they would imply that Hitler himself suffered from some or other mental illness.

Luckily, they didn't do so. Which brings us to the central issue of your question. Even if a family history of mental illness makes it more likely for an individual to suffer from mental illness, the correlation itself is not evidence for the existence of mental illness in the individual.

For example, much has been written about Friedrich Nietzsche's illness. An interesting discussion for erudite, history-minded psychiatrists and doctors, no doubt. However, how useful is the question to the historian? What would a specific illness tell us about the events of history that a non-descript one would? Often very little.

I'm not really familiar with Thomas Jefferson, but let's use him to address the more dangerous point of retrospective diagnosis gone too far. Let's assume he was autistic. Alright. Things still happened as they did. It might explain some possible choices not taken (a viable study of historical enquiry), but it does not change the historical outcomes themselves. Such inquiries seem to invite the counterfactuals historians only care for at the dinner table or in the bar:

"Thomas Jefferson was autistic. What if he wasn't?"

To reiterate. 1) Retrospective diagnosis is methodologically shaky. Testimony about some historical figure's mental state is, at best, circumstantial. 2) It does not really matter for historical inquiry. Lenin's strokes are historically significant because they indirectly resulted in the reconfiguration of Soviet politics. The pathology of his strokes are historically, broadly speaking, insignificant.

So this isn't the perspective of a historian exactly, but I've discussed this topic with my dad, who's a doctor, specifically a type that does a lot of diagnosing (pathology).

The issue inherent in this becomes apparent when you think about how challenging diagnosing people IRL is. We, today, can ask the patient questions, document symptoms, and perform texts. But even with those tools at our disposal doctors can be unsure of diagnoses, which is why people get second opinions. With retrospective diagnoses we're asking people to diagnose based on whatever records are available. We can't do the usual diagnosing process, which makes it methodologically suspect.

An example of this is in diagnosing Lincoln with Marfan's. Marfan's has the rather obvious symptom of causing people to be unusually tall. This is tempting, since Lincoln was noted for his height for for being rather slender. Lincoln also had long limbs, which is another symptom of Marfan's. However, there's a lot more to diagnosing Marfan's than that. Doctors use family histories and a variety of physical characteristics. We can't ask Abe if his parents had those symptoms, particularly since the condition wasn't documented until after his lifetime. We certainly can't have him examined easily by a doctor. And ultimately, since Lincoln didn't have many other characteristics of it (particularly vascular problems), it's rather unlikely--he probably was just tall. This brings it to another problem. It's extremely tempting to chalk things up to interesting syndromes and conditions. But without any way of disproving them in many cases, it's very subject to bias. If you're looking to make a diagnosis of something, you're probably going to find some evidence for it. But that's not how medicine really works.