All sensor observations are stored in the HA database after same principles as all other numeric measurements.
You can use the statistics integrations to create the most common aggregations. They do some data reduction to keep the database from growing too much, but I think it is mostly OK.

I think it out of scope to mimic these statistics inside this integration.

Cool. Appreciate the feedback and I will have a look at using the statistics integration for this purpose. The previous v1 integration did include the data, but only because it was supplied with every response. I was hoping to be able to pull in my old historical high/low data from v1 into v2, but that might take some database magic at this point. Thanks again. Your integration works great.

Messed around with it a bit and it appears that the statistics integration does not support "month" or "year". I was able to do 30 days and 365 days, but no idea how those will work out yet.

While it may be out of scope, it does seem unfortunate that the integration does not provide the high and low temperatures for the current day. Those seem like fairly basic weather statistics. I would say that they are easy enough to replicate with the statistics integration, except the statistic integration doesn't seem to have any concept of "today". It only seems possible to do high/low temps for the past 24 hours or 1 days - not "today" or "current day" apparently.

Kind of ironic how they double-down on rain stats, but nothing on anything else. Oh well, I'll figure something out.

Agree that it is a pity that the daily highs and lows are absent from API V2. The calculations are obviously carried out as the data is available in the Davis Weatherlink app.
What I meant by 'out of scope' is that my aim was to present data from the API and refrain from making calculations and other processing in the integration. It is possible to replicate parts the statistics integration but that would probably a considerable maintenance burden.

I must admit that I have not looked deep into the statistics integration but from a quick glance at the documentation it might be possible to combine template features to create the parameters for first and last timestamps to include in the calculations.

The concepts of today, yesterday, this month etc are present in the energy dashboard so someone have had the same thoughts.

If you or anyone else is willing to create a PR or at least a proof-of-concept here I will be happy to review and consider it. But I think it would be more resource efficient to add today, this_month, this_year etc as parameters in the statistics integration.