Weight random Age choices like Legacy and expand regression coverage (#41)

docs/GAMETESTS.md

@@ -47,6 +47,27 @@ automated check, and hand its exact scenario to the player for in-game
 verification. Do not keep retrying a stalled test or leave its process running.
 Keep the narrower non-stalling plan, persistence, and runtime checks automated.
 
+For issue #41, the combined archive and the focused `linking` archive initially
+stalled during the multi-profile travel acceptance test. That test is no longer
+registered with GameTest; the rerun passed all 116 registered tests before the
+subsequent random-biome-count adjustment. The 51-test core suite passes after
+that adjustment, but the full archive stalled again late in its first batch;
+do not treat that interrupted run as a pass. Verify its
+scenario manually: in a fresh world, bind
+one Descriptive Book for each of Standard, Flat, Amplified, Island, Cave, and
+Void terrain (use the corresponding Terrain Page to select each one). Visit
+each Age, confirm a safe arrival platform and headroom, and
+travel back using a bound Linking Book. The narrower creation, persistence,
+terrain, and travel checks remain automated. See
+[random Age weights](RANDOM_AGE_WEIGHTS.md) for the Legacy rank mapping and
+the opt-in weighting regression.
+
+Also bind several Link-Panel-only Descriptive Books with no authored biome
+pages. Check that each Age retains its biome set after leaving, unloading, and
+reopening it, and that the selected biome count varies between Ages without
+exceeding the eligible pool. This covers the bounded random-biome path touched
+after the last successful full-archive run.
+
 The optional `-PwithSereneSeasons=true` runtime is for a real dev client, not
 the GameTest server: Serene Seasons sends a season payload to GameTest mock
 players, which cannot negotiate its channel. Follow the real-client Serene test in