Did you really not have any forwarded queries?
I'm not on the bleeding edge here, but I'm seeing the same thing, although not quite as bad.
$ pihole -v
Pi-hole version is release/v5.4 v5.3.1-94-ge163ed44 (Latest: v5.3.1)
AdminLTE version is release/v5.6 v5.5.1-57-geb4b88d (Latest: v5.5.1)
FTL version is release/v5.9 vDev-b90a993 (Latest: v5.8.1)
Pie chart after running Pi-hole for 3 days:
Total = 95,7%
Pie chart immediately after restarting the DNS resolver:
Total = 100%
Importing from the database seems to work correctly, but during prolonged uptime some queries seem to not be counted (or over-corrected).
Noteworthy here is that the values particularly for localhost#5335 (unbound) changed quite significantly. But also my target for reverse lookups, fritz.box#53, has increased by almost a third of the previous value.
There was a similar issue a while ago, but the total back then was too high. Now it is too low. My blind guess would have been that maybe the internal queries from pi.hole would not have been counted, but those make up for only 2% of all queries in my case, which is not enough to explain the 4.3% that were actually missing. Otherwise I don't really have any immediate ideas about what the issue could be.