As I've mentioned before, mentioning time frames as facts (through, it should be said, nothing more or less than being a concise answer) does not
turn them into facts
if the context they are given in is one of "the release date we've initially provided is a target",
especially if they're a reiteration of said date. You can't expect any question of "When does 1.3 come out?" (which there are
a lot of) to be answered with "Q2 2017 at the earliest, it could be later but we don't expect it to be." every single time, as opposed to simply saying "Q2 2017", just because you want to reiterate a previously stated point (the release date only being a target) ad nauseum.
Imagine if we did that with the weather. We'd start suing meteorologists for 'promising' certain weather and then failing to deliver whenever they say "Tomorrow it's going to be sunny." instead of "Tomorrow it is probably going to be sunny, but it may not be because this is only a prediction.". We don't, however, because we
know that even though weather forecasts are
presented as facts, they are provided in a context of uncertainty. It's exactly the same thing with release estimates.