Sample the prefix-length pages with bounded lookups, each prefix once (closes #66)

/prefixlength/{length} and /prefixlength6/{length} grouped every live route
of that length by prefix, shuffled all the prefixes and joined the routes
back, so a request grew with the number of routes and listed a prefix once
for each peer that shared its newest time.

Now a length with at most the page's 500 prefixes has all of them listed. A
longer one is sampled: each lookup draws a random row id and takes the next
route of that length from the mask length index, at most four lookups per
prefix asked for. Each picked prefix is shown once, with its newest route.

Model: opus-5-5

TODO.md

@@ -27,6 +27,11 @@ real 5 GiB limit (https://git.eeqj.de/sneak/routewatch/issues/3).
 
 # Completed Steps
 
+- 2026-10-14: `/prefixlength/{length}` and `/prefixlength6/{length}` no
+  longer read every route of that length: a length with at most 500
+  prefixes has all of them listed, a longer one is sampled with random
+  lookups on the mask length index, and each prefix is listed once, with
+  its newest route (closes #66)
 - 2026-10-03: `/api/v1/stats` serves the prefix distribution from memory,
   seeded at startup and adjusted on every live-route write, so a request no
   longer reads every live route (closes #30)