Define "kill a feature".
Few companies have "one feature" to do and then decide to do it or kill it.
So "create new product X" or "add new feature Y" requires a priority decision, where the amount of resources becomes important.
It's obvious no bright manager lets the R&D resources just sit and idle when there exists a list of potential tasks to do. But many R&D managers are likely to be in a situation where one of their top requests is to get more staff. And many project managers are wanting more staff assigned to their projects.
That's different. I can see shelving something to prioritize something else, that happens all the time. I can see killing a feature because it can't be done cost effectively or can't be marketed. But what the ML folks do is not controversial and would do nothing but add value to Canon cameras and at very little cost. Heck, they could just sub-contract the ML team in incorporate their stuff today at almost no cost.