I've been told (but I don't understand) that it's a good idea to use a number of chess engine to analyze a chess opening. It this true? Why is this a good idea?)
Chess engines use heuristics, i.e. they look at a position several moves ahead, and quantitatively evaluate the position using some criteria (the most straightforward example is to just count the pawns and pieces). The move that leads to the position with the highest score, assuming best play (according to the engine's heuristics) from both sides is the engine's best move.
So the evaluation of the position depends on the choice of heuristics. Different types of positions are best described by different heuristics. Different engines use different heuristics. Therefore, in an opening where several types of positions could arise, different engines may have different opinions on what the best move is. (I haven't tried this myself, but I'd bet the top 5 chess engines won't agree on 1.e4, 1.d4 or 1.c4)
I think that is someone's opinion. A few years ago (maybe more than few) I routinely beat some of the commercial chess engines. They are much better today, but facing a (skilled) human opponent is much more helpful. Although now a days, you're sure to get you butt handed to you if you play a quality engine on the most advanced levels.
If you have no human opponent to study , you have to choose the engines . But Firstly, you have to recognize the engines weakness and strengths . Like if you analyse some routine openings , it will be a bad decision. A complicated position or something like that needs the engine. Secondly, Some engines also fail to find a clear path in the complex middle game or end games . Like I tried to analyse a game of Dragon variation(Sicilian defense) But engine failed to find immediately a right path(with Houdini 4 and Komodo 6). So, you can study with two engines ( not more than that because that's not going to helpful for your study) , but it sometimes never helpful .