web: sim store, URL state scheme, debounced parallel runs (M3 slice 2)

useSimStore is the one reactive module singleton from the spec: panels
over a shared base config, runs debounced 400 ms, one POST /api/scenario
per panel in parallel, results swapped in atomically only when every
request succeeds. Failures keep the last good results; engine 400s are
routed to an inline validationError, everything else to the banner
message. Stale in-flight responses lose to newer runs per panel.

URL state follows spec section 7 (readable params, repeated panel=,
focus index, full fallback to the preset on any malformed part) and is
covered by round-trip and rejection tests. The study copy and initial
panels live in the deepfake-school preset module; formatPct is the
single percent-rounding rule.

The vitest tsconfig now keeps lib at ES2022 like the app config, and
prettier skips the tygo-generated src/types (drift guard stays green).

web/src/presets/deepfake-school.ts

@@ -0,0 +1,49 @@
+import { StrategyMostConnected, StrategyNone, StrategyRandom } from '@/types/engine'
+import type { StudyPreset } from './types'
+
+// The one study this deployment tells. A different research question is a
+// new preset file plus a changed import in the app; no component changes.
+
+export const deepfakeSchoolPreset: StudyPreset = {
+  headline: 'How a non‑consensual deepfake spreads through a school',
+  narrative:
+    'One simulated year group, 120 students. Educate the same 30% of them, ' +
+    'but change who: the fake reaches {0} of the school with no program, ' +
+    '{1} educating at random, and {2} educating the best‑connected students.',
+  disclaimerShort: 'Illustrative model, not validated',
+  disclaimerLong:
+    'spreadlab runs a seeded agent-based toy simulation: a synthetic ' +
+    'friendship network of one school year group, a fixed chance to forward ' +
+    'the fake along each friendship per round, and an education program that ' +
+    'teaches some students to refuse. Its parameters are chosen for ' +
+    'illustration, not fitted to data, so it is not a validated prediction ' +
+    'of any real school. Use it to build intuition about who to educate, ' +
+    'not to forecast outcomes.',
+  toneThresholdPct: 30,
+  base: {
+    numStudents: 120,
+    edgesPerNode: 3,
+    triangleProb: 0.45,
+    forwardProb: 0.38,
+    numEducated: 36,
+    origin: 0,
+    graphSeed: 17,
+    thresholdSeed: 2,
+    educationSeed: 1,
+  },
+  panels: [
+    { id: 'preset-no-program', label: 'No program', strategy: StrategyNone, overrides: {} },
+    {
+      id: 'preset-random',
+      label: 'Educate 30% at random',
+      strategy: StrategyRandom,
+      overrides: {},
+    },
+    {
+      id: 'preset-most-connected',
+      label: 'Educate best-connected 30%',
+      strategy: StrategyMostConnected,
+      overrides: {},
+    },
+  ],
+}

web/src/presets/types.ts

@@ -0,0 +1,26 @@
+import type { Config, Strategy } from '@/types/engine'
+
+// Frontend-only presentation types (spec section 1). They describe study
+// copy and the initial scenario set, not simulation data: Config, Result
+// and Strategy stay imported from the generated types.
+
+export interface PanelSpec {
+  id: string // stable client id
+  label: string // user editable
+  strategy: Strategy
+  overrides: Partial<Config> // sparse diff against the base config
+}
+
+export interface StudyPreset {
+  headline: string // hero h1
+  // Hero paragraph; {0} {1} {2}... placeholders are replaced by each
+  // panel's reached%, formatted by the shared formatPct.
+  narrative: string
+  disclaimerShort: string // badge text
+  disclaimerLong: string // About popover body
+  toneThresholdPct: number // reached% <= threshold renders "good" (teal)
+  base: Config // the default world
+  panels: PanelSpec[] // initial panels
+}
+
+export const MAX_PANELS = 6