Drop the stale-read workaround; the cache is gone

Preferring the event's figure over the read existed only because the catalog
gateway cached reads for 60 seconds. That cache has been removed at the source,
so the read is current and there is no second source of truth to arbitrate.

The crossing context is now appended only while the reference is still below
the threshold — an entry outlives the condition it describes.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

workflows/README.md

@@ -108,32 +108,37 @@ webhook — its data lives in that workflow and there is no other way in. The no
 is `onError: continueRegularOutput`: an availability answer must not depend on
 the watchlist being up.
 
-### A dated fact beats an undated read
+### A stale read once made it contradict itself
 
-The two figures can disagree, and when they do the **read** is the wrong one.
-This was not theoretical — the agent answered, in one sentence:
+The agent published this, in one sentence:
 
 ```
 "68 in stock, above the reorder threshold of 50 … It crossed the threshold, from 68."
 ```
 
-The catalog domain caches reads for **60 seconds**. An agent woken by an event
-reacts in seconds and then reads a figure taken up to a minute earlier. A
-movement is the catalog's own *dated* statement that stock went from X to Y; a
-read is an undated copy. So the movement wins, and the answer says the basis
-changed:
+Both halves were internally consistent and the answer was wrong. The catalog
+domain was caching reads for **60 seconds** at the gateway, so an agent woken by
+an event read the figure from before the event.
 
-```
-Basis: the latest stock movement, which is more recent than the catalog read
-```
+The first fix was in this workflow — prefer the event's dated figure over the
+undated read. It has been **removed**, because the cache was removed instead:
+caching a read-write API buys little and costs correctness on every mutable
+resource, since the gateway has no invalidation on write. Working around it here
+would have left four other domains lying to whatever reads them next.
 
-Measured after the fix, with the cache deliberately warmed at 38 and the true
-stock moved to 8 inside the window: the answer said **8**.
+Measured after: a `PATCH` to 300 followed three seconds later by the same
+question answers **300**.
 
-This only helps a reference the reactive half is already watching. For any
-other, the read stands with whatever staleness it has, and nothing in the
-workflow can tell — which is worth remembering before treating any single read
-as current.
+The lesson worth keeping is not about caches. **Do not treat a single read as
+current unless something guarantees it** — and an agent cannot tell.
+
+### The crossing is mentioned only while it still holds
+
+An entry stays on the watchlist until a human clears it, so a reference that has
+since been restocked still has one. Saying "300 in stock … it crossed the
+threshold, from 68" reads as a contradiction even though both halves are true,
+so the context is appended only while the reference is actually below the
+threshold.
 
 ## approvisionnement-periodic-reorder