Why was this counted blunder?

Excellent and best should have been close, but ..

You should try a short course in programming :-)

On the net you have a lot of free resources for learning python for example. Then you would understand why those kind of thing occurs and the limit of coding in those situation. 

I mean coding complex situation like this is horrible because you literally cannot think of every situation.

Like here you have a huge score gape between excellent and best, but if the ranking is not based solely on ranking those kind of situation arise. 

For example each time I had a great move, I had the message " you find the only good move in this position ". Meaning great move does not have only in increase in score but apparently it has to be the only "good move ". 

That's one of the many issue that can arise :-)

Right.  Not only should Excellent and Best be close but Excellent should also be excellent!

Actually I spent more than 3 years programming. I first started in scratch for 2 years (I was even able to make 3D games there where it was recognized impossible). Then, I spent 1 year in Python, Arduino, and GMS2

You should try a short course in programming :-)

On the net you have a lot of free resources for learning python for example. Then you would understand why those kind of thing occurs and the limit of coding in those situation. 

I mean coding complex situation like this is horrible because you literally cannot think of every situation.

Like here you have a huge score gape between excellent and best, but if the ranking is not based solely on ranking those kind of situation arise. 

For example each time I had a great move, I had the message " you find the only good move in this position ". Meaning great move does not have only in increase in score but apparently it has to be the only "good move ". 

That's one of the many issue that can arise :-)

Ok so look at it like a coding problem. 

You have data on the players, data on the opening (because some of the ranking will be based on known opening theory) and you have stockefish evaluation. Given this you have to rank each move. 

I am not saying that it is impossible to do, but given the fact that the engine are far from being perfect, how do you want a programming layer on something so complex to translate to something else then even less perfect ? 

Of course you will have software blunder, a lot .