Applebaum was an expert witness called in to testify on how a computer could be remote controlled, and the prosecutor tried to have him dismissed based on the retweet. He ended up testifying, so I guess the attempt wasn't successful.
Assuming a retweet is the full extent of their "friendship", this is the real-world equivalent of dismissing an expert witness because the defendant quoted them in a paper once.
Something like a tweet is context dependent. Some people tweet mostly to their friends and a lot of retweets are in fact indicative of a personal relationship. It will often be easy to estimate for a twitter user, but very hard to make objective enough to work as evidence in court.
Do they have just 12 or 12000 followers? Are they tweeting personal tidbits or politics & jokes? etc. These things can add up to an very informed guess.
A retweet is certainly relevant. The evidence makes it more likely that two people are friends than the baseline where there is no retweet.
[1] http://en.m.wikipedia.org/wiki/Evidence_under_Bayes_theorem