I remember being on Twitter on most of election day and seeing evidence of all sorts of manufactured rhetoric from Reuters, and was really disturbed (proof below). There are a lot of other explanations for why this happened, but I'm not inclined to believe them given the reality of what happened on election day. It looked like Reuters' polling towards the 8th showed Trump taking the lead, but they deleted the data an…
First, they reported on early election numbers in North Carolina which directly contradicted their poll numbers and showed Trump performing better than predicted.
Second, they outlined a clear and realistic path to Trump winning. While their own poll numbers did not support this fact, they ultimately included it. They also did not rule out a Trump win in any way.
It is always important to note that a 90% chance of winning is in no way a guarantee, and I do not believe Reuters ever truly ruled out Trump as a contender. They used the only early data available, polls--and when every poll is wrong, and psychics are still not a real thing, they did what they could with the data they had. At least, in my reading of the article you linked.
Regarding the polls you linked, those are not the Reuters/Ipsos polls of 15,000 participants. Those are much smaller polls (850 users) with far larger margins for error.
While I cannot speak to the censoring of the 11/6 poll mentioned in the second tweet, having Trump leading in popular vote nationally (eg: what polls generally check) was ultimately incorrect, regardless of what you believe. Polls need to be carefully read, interpreted, and corrected for -- which is why fivethirtyeight gave Trump a much higher chance to win than any other body.