ci: перекалибровать общий потолок spaceSwitchMs изометрии по ряду CI (#675)

spaceSwitchMs изометрии — одно холодное переключение этажа со сборкой
2.5D-геометрии; он растёт с каждой стадией 2.5D, и потолок 1800 мс из #89
stage 1 стал самим уровнем. Медиана perf-smoke на hosted-раннере:
1500 мс (09-09…09-12, 12 прогонов) → 1668 (09-19…09-25, 6) → 1798 после
#649 (9, σ 42, максимум 1867,7). Попытки одного SHA расходятся на 0,4 %,
разные прогоны одного SHA — до 5,5 %: больше образцов вердикт не меняют,
рычаг — потолок.

2200 мс: +17,8 % над наблюдённым максимумом и ниже единственной настоящей
регрессии окна (2719,6 мс, #583 до решётки, 09-16); удвоение уровня
краснеет с запасом. Одно число в трёх файлах: смок = полный профиль
(#473 AC4), плотный двойник = исторический (#160). Коэффициент и допуск
полного сравнения не тронуты. Обоснование — demo/performance/README.md,
тест закрепляет число и три точки ряда.

Issue: #675
User-Visible: no
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018qZfe7YS4rqEMKoVeS3GKd

demo/performance/README.md

@@ -99,6 +99,22 @@ still below the known failed pre-optimization result (3501.5 ms). This aggregate
 is a catastrophic guard; the full workflow's base-relative comparison remains
 the detector for smaller regressions.
 
+The isometric `spaceSwitchMs` ceiling is 2200 ms, shared by the smoke, the full
+isometric profile and its Stage 3 dense twin (#675). The metric is one cold
+floor switch that builds the second floor's 2.5D geometry, so it grows with
+every 2.5D stage, and the original 1800 ms from #89 stage 1 had become the level
+itself. The hosted-runner smoke median rose from 1500 ms (2026-09-09 to 09-12,
+12 runs) to 1668 ms (09-19 to 09-25, 6 runs) and 1798 ms after #649 (9 runs, σ
+42 ms, maximum 1867.7 ms). The median inside a run is already stable (re-run
+attempts of one SHA differ by 0.4 %), separate runs of one SHA differ by up to
+5.5 %, and the level itself moved by 20 %: more samples would not change the
+verdict, so the ceiling is the lever. 2200 ms leaves 17.8 % over that maximum
+and stays below the only real isometric regression of the window, 2719.6 ms
+(#583 before its lattice fix, 2026-09-16); doubling today's level fails by a
+wide margin. Occasional hosted runners are about a quarter faster, which a
+ceiling does not mind. Smaller growth stays the job of the full workflow's
+base-relative comparison.
+
 The dedicated `performance.yml` workflow is the full comparison. It runs on
 every `main` promotion, weekly and on manual dispatch for an important beta or
 performance-sensitive change. It checks out the candidate and its base SHA,

demo/performance/budgets-isometric-smoke.json

@@ -17,7 +17,7 @@
     },
     "spaceSwitchMs": {
       "stat": "median",
-      "hardMaxMs": 1800
+      "hardMaxMs": 2200
     },
     "stateUpdateMs": {
       "stat": "median",

demo/performance/budgets-isometric-stage3-dense.json

@@ -8,7 +8,7 @@
     "viewToggleMs": { "stat": "median", "maxRegressionRatio": 0.2, "noiseAllowanceMs": 100, "hardMaxMs": 1500 },
     "openingUpdateMs": { "stat": "median", "maxRegressionRatio": 0.2, "noiseAllowanceMs": 75, "hardMaxMs": 1000 },
     "overlayInteractionMs": { "stat": "median", "maxRegressionRatio": 0.2, "noiseAllowanceMs": 75, "hardMaxMs": 1000 },
-    "spaceSwitchMs": { "stat": "median", "maxRegressionRatio": 0.2, "noiseAllowanceMs": 100, "hardMaxMs": 1800 },
+    "spaceSwitchMs": { "stat": "median", "maxRegressionRatio": 0.2, "noiseAllowanceMs": 100, "hardMaxMs": 2200 },
     "stateUpdateMs": { "stat": "median", "maxRegressionRatio": 0.2, "noiseAllowanceMs": 75, "hardMaxMs": 1000 },
     "resizePreviewMs": { "stat": "median", "maxRegressionRatio": 0.2, "noiseAllowanceMs": 150, "hardMaxMs": 2200 },
     "panZoomMs": { "stat": "median", "maxRegressionRatio": 0.2, "noiseAllowanceMs": 60, "hardMaxMs": 600 },

demo/performance/budgets-large-house-isometric.json

@@ -6,7 +6,7 @@
     "modelReadyMs": { "stat": "median", "maxRegressionRatio": 0.2, "noiseAllowanceMs": 200, "hardMaxMs": 3000 },
     "firstStableRenderMs": { "stat": "median", "maxRegressionRatio": 0.2, "noiseAllowanceMs": 250, "hardMaxMs": 3500 },
     "viewToggleMs": { "stat": "median", "maxRegressionRatio": 0.2, "noiseAllowanceMs": 100, "hardMaxMs": 1500 },
-    "spaceSwitchMs": { "stat": "median", "maxRegressionRatio": 0.2, "noiseAllowanceMs": 100, "hardMaxMs": 1800 },
+    "spaceSwitchMs": { "stat": "median", "maxRegressionRatio": 0.2, "noiseAllowanceMs": 100, "hardMaxMs": 2200 },
     "stateUpdateMs": { "stat": "median", "maxRegressionRatio": 0.2, "noiseAllowanceMs": 75, "hardMaxMs": 1000 },
     "resizePreviewMs": { "stat": "median", "maxRegressionRatio": 0.2, "noiseAllowanceMs": 150, "hardMaxMs": 2200 },
     "panZoomMs": { "stat": "median", "maxRegressionRatio": 0.2, "noiseAllowanceMs": 60, "hardMaxMs": 600 },