docs: explain client reuse and replacement for waiting pool callers

docs/pages/features/pooling.mdx

@@ -56,6 +56,14 @@ client.release()
   future requests to check out a client from the pool will wait forever.
 </Alert>
 
+With `max: 1`, a second caller waits while the only client is checked out. How the first caller releases that client determines whether the waiting caller receives the same client or a replacement:
+
+![Caller A holds the sole pool client while B waits. When A calls release without an error, the healthy client is reused and B receives the same client. When A passes an error to release, the pool removes that client and connects a replacement for B.](/images/pool-release-outcomes.png)
+
+This example assumes an active pool, a successful replacement connection, no connection timeout, and no lifetime or usage limit that removes an otherwise healthy client. Passing an error to `client.release(error)` discards that client; the figure describes this explicit release choice, rather than prescribing it for every query error. The replacement still needs to be released after use. `max: 1` makes the example easy to follow, rather than recommending a pool size.
+
+[Download editable SVG](/images/pool-release-outcomes.svg).
+
 ### Single query
 
 If you don't need a transaction or you just need to run a single query, the pool has a convenience method to run a query on any available client in the pool. This is the preferred way to query with node-postgres if you can as it removes the risk of leaking a client.