Give the reason 0105's queue answers are unwritten that survives the queue

src/cache/envelope.rs

@@ -43,8 +43,22 @@
 //! three different answers: a drop is reported at `failure` rather than
 //! `notice`, a drain steps over a bad entry instead of stopping at it, and a
 //! counter that fails its own envelope is rebuilt rather than emptying the
-//! queue. Nothing in this tree holds a queue, so none of the three is written
-//! here, and #47 is where they belong.
+//! queue. None of the three is written here, and #47 is where they belong.
+//!
+//! What holds all three is that nothing puts a queue entry in an envelope at
+//! all. [`crate::server::write_queue`] keeps the queue in memory for as long as
+//! the value lives and says of itself that its durability is unbuilt and that no
+//! drain runs, so there is no stored queue entry for one of these checks to
+//! refuse and no drain for one of them to continue past.
+//!
+//! THIS PARAGRAPH GAVE A DIFFERENT REASON: THAT NOTHING IN THIS TREE HOLDS A
+//! QUEUE. That was the tree on 2026-08-28, when this section was written. The
+//! queue arrived two days later, in a change under `src/server/` that had no
+//! reason to open a `What is not here` section in the cache, and the module says
+//! in its own words that it holds the queue a drain would walk. The conclusion
+//! did not move and the reason did, which is why the reason is replaced with the
+//! one that survives the queue arriving rather than corrected to a second one
+//! that would not. #412 is where it was found.
 
 use core::sync::atomic::{AtomicU64, Ordering};