Yup, you're right. I just skimmed through the original article, but looking closer now it does look like they just averaged all the months then compared those numbers. So seasonality is NOT taken into account.
To do some offhand numbers on this real quick. Monthly death averages are pretty hard to find, at least for several years in a single chart/place. This old reddit post has a decent chart (huge, multi-image in one or I'd link it) but only goes through end of April, and this CDC page is average dailies for just 2017. Both graphs look similar.
The original post data went through week 32, or mid-August. From several other articles, the deadliest months are Jan, Feb, and Dec. But, Sept/Oct/Nov, the remaining other months for this year, are all below the yearly average. Squinty-eyed estimating from the CDC graph, it looks like the seasonal high in December is offset by the seasonal lows in the other three remaining months, to where the rest of the year is about average in total. So, to rough out the article's data, it looks like they could assume the rest of the year is average, to equitably compare this year to previous years. So seasonality should NOT affect the data much, and their comparison data should be pretty close to final results, unless deaths suddenly go crazy high or crazy low the rest of 2020.
--------------
Or, they could have simply looked at every year up until week 32, then compared those death counts. No seasonality or any other issues there. Straight 1:1 comparison... No clue why they wouldn't have done that in the first place. You'd think looking at proper methodology would be pretty early on the process for Dr Huber... Not after publication..