fix(bench): reset instance JVM settings for sessions without --jvm

apply_jvm wrote instance.cfg and never restored it, so B5-B12a in run 1
ran on B4's last set (Shenandoah, 6 GB). Within-block verdicts hold
(same JVM on both sides); B7's no-shader baseline is re-run in run 2.

Also: B14 spawn-count automation (b14/: fresh seeded world per run,
offline-located biome sites, datapack entity counts, journaled config
variants) and RESULTS-proxy.md with run-1 verdicts.

analysis/client-perf/harness/b14/sites.json

@@ -0,0 +1,49 @@
+{
+ "source_zip": "/home/logan/onion-bench/worlds/bench-world-v1.zip",
+ "source_sha256": "144caf5fcdfea1a0eac0c787fdcded949e1a67df6ac112400f37be30ce7fb76e",
+ "seed": -306676217,
+ "avoid": [
+  [
+   0,
+   0
+  ],
+  [
+   3000,
+   -9
+  ]
+ ],
+ "stride": 1,
+ "note": "y = first free block (full chunks: MOTION_BLOCKING_NO_LEAVES, proto chunks: WORLD_SURFACE_WG); the player stands on a barrier at y-1. Valid only for this seed + mod set (biomes are seed-deterministic).",
+ "sites": {
+  "river": {
+   "x": 904,
+   "y": 63,
+   "z": 88,
+   "biome": "minecraft:river",
+   "frac": 0.373,
+   "score": 0.373,
+   "full_chunk": false,
+   "candidates": 17
+  },
+  "jungle": {
+   "x": -8,
+   "y": 107,
+   "z": -201,
+   "biome": "minecraft:jungle",
+   "frac": 0.52,
+   "score": 0.52,
+   "full_chunk": false,
+   "candidates": 19
+  },
+  "warm_ocean": {
+   "x": -200,
+   "y": 63,
+   "z": 56,
+   "biome": "minecraft:warm_ocean",
+   "frac": 0.75,
+   "score": 0.75,
+   "full_chunk": false,
+   "candidates": 31
+  }
+ }
+}