I'd say you're reading it correctly. This is what I see:
There's an underlying yearly cycle that goes something like:
- gets hotter around March
- gets colder around May-June
- gets hotter around August
- gets colder around November
This is, let's say, a sort of cyclic baseline that repeats every year. But over it there are variations. Some years are generally colder, or generally hotter. Some the differences are bigger or smaller, etc. Then, obviously, it's not a perfect curve and there's "noise" on it. It's also interesting to note that the "March peak" is generally hotter than the "August peak", and that the "November valley" is generally colder than the "June valley". All this, broadly speaking.
Now, until last year, the highest temperature had happened in 2016, in the "March peak". This "makes sense". I mean, it makes sense that the hottest it got was on one of the hot peaks. What's more, it happened in the hotter one of both hot peaks.
But if we look at this year, what we see is two things. One is that, as the title points out, there's been a new highest, higher than March 2016. This is pointed out because it makes an easy and simple headline.
The second thing you can see is what you pointed out, that in 2023 we haven't had much of a valley in May-June and instead it has been hotter since March producing that the August peak be not only much higher than other years but even higher than the highest March peak. (And, I may add, it may still get hotter in this peak)