Something to ponder and maybe worth some input from Tom.
I think the problem here is a timing issue between the time a report is being generated on one WeeWX instance and the time that records are added to (at least) two different databases. A key point to remember is that for $alltime to use the daily summaries the timespan concerned must start and end on either the first timestamp in the archive, a midnight boundary or the last timestamp in the archive. If this condition is not met the archive is used instead of the daily summaries.
The use of two different bindings implies two different databases. Whilst I don’t think we have been given a detailed description of the setup, I assume there are at least two WeeWX instances running; one with a vantage station and the other a Weatherflow each using their own database. The vantage station will be saving an archive record to database almost immediately after the end of the archive period plus archive delay. I am not familiar with the Weatherflow station, but it is quite possible the archive records are not generated until sometime after the end of the archive period plus delay, typically on receipt of the next loop packet. In such cases, if the vantage instance produces a report using the Weatherflow binding (database) the timespan used by the alltime tag will likely not match the first and last timestamps of the Weatherflow database meaning the archive will be used rather than the daily summaries. However, when the same tag (with the Weatherflow binding of course) is used on the Weatherflow system the timespan in use covers the first and last timestamps in the Weatherflow database (because the reports are run until after the archive record has been saved) so the daily summaries are used. Hence the different results for the seemingly same tag on the two different systems.
Should be easy enough to confirm by comparing the logs from the two instances.
How to fix? I’m not sure, typically the way timing differences between reports and multiple databases are handled is to use $latest in place of $current, but that won’t work here. I guess you could design some elaborate system of reports where the Weatherflow system generates/updates a file that the vantage system then uses - sounds like a lot of work. Or you could live with it. Worth noting though that you do pay a performance penalty by doing a query over the entire archive, so if you do live with it as is I would keep any eye on your report times and not go adding numerous more similar tags.
One out of the box solution could be to increase the vantage system archive delay so that reports are not run before the Weatherflow system has saved its archive record. But this approach may have other undesirable effects, eg delaying the production of vantage reports/pages.
Then of course Tom may have a magic solution.
Gary