Suggestions are never as black and white as that. There's enough uncertainty on your sales to prove that 16 flat won't cut it, so let me break it down.

By excluding 15, you are excluding (potentially) four sales below 16, all in favor of two non-bot buyers, which notably are not guaranteed to keep their buy orders active, or even go through with the buy orders and break rules by declining. That's why while they're not invalid, they're not considered to be as strong counter-proof as the bot buyers would be.

On the other end, there's no actual counter-proof for 18, besides the existance of the other low sales. Your point would be more valid if the high sale here was more discrepant size-wise, but that's not the case here. Not to mention, two of the possible sales here could either counter 18, or support 18 (either they're 16/20, or 8.5/20).

The difference between going with 16 flat as opposed to 15-18, is that with a flat price, you're ignoring a vast number of sales in favor of one (confirmed) sale, while 15-18 captures a bigger spectrum of sales to account for the uncertainty being presented here. 15 on it's own is too low, and 18 on it's own is too high, but as a range they make up for eachother for a median of 16.5.

You would have a much stronger case here if the proof for 16 was stronger. Be it by more buyers at 15-16 (bots or otherwise) to counter 15, sellers at 17-18 to counter 18, and (CONFIRMED) sales to strengthen 16 as it's own price point. But none of it is present here, so I believe it's better here to account for the uncertainty and go for a range. Since, if any of the confirmed sales were to be confirmed (or the buyers and sellers changed overnight), they could shift this suggestion in a completely different direction. Going with 15-18 is the best way to account for any possible change in this specific suggestion.

I should add that most suggestions don't have to take this extreme direction. Most suggestions ARE as simple as "buyers above sale or sellers below, excluding". But it's never been a hard rule, and it can be overturned if the sales presented are substantial enough to make up for the counterpoints. You have to analyze all of the sales you've gathered (even the invalid ones) and come up with the best possible conclusion.