feat(sim): sim_radio_get_params export + fix identity-collision RNG bug

meshcore-solo-site's new cross-visitor relay bridge (Phase 3) needs a way
to read a sim instance's live freq/bw/sf/cr so a WebSocket relay server
can fan traffic out only to other visitors tuned to the same params,
mirroring real LoRa channel isolation. Added sim_radio_get_params()
(examples/companion_radio/main.cpp), reading NodePrefs directly -- the
real live source of truth, since SimRadio::setParams() has always been a
complete no-op (accepts and discards its arguments). Out-params via
pointer, same buffer-pointer idiom sim_radio_poll_tx() already uses;
build_wasm.sh needed HEAPF32/HEAP32 added to EXPORTED_RUNTIME_METHODS so
JS can read them back.

While testing the relay bridge with two independent browser contexts, hit
a real bug: sim_instance_salt() (SimInstance.h) is a pure function of the
simInstanceTag STRING ('hero'/'B'/'R'), so it's only useful for telling
apart same-tab instances -- it's identical for two genuinely different
visitors who both boot a 'hero' instance. Combined with SimRNG::begin()'s
other seed ingredients -- time(NULL) (1-second resolution) and a WASM
heap pointer (no ASLR inside the sandbox, so it's fully deterministic
across independent boots of the same binary) -- two different visitors
landing on the same wall-clock second reliably generated byte-identical
Ed25519 identities (confirmed: two fresh browser contexts launched
together produced provably identical advert packets end to end).

Fixed by mixing in real crypto.getRandomValues()-sourced entropy from the
host page (new sim_instance_entropy(), read the same way simInstanceTag
already is) into both SimRNG::begin() and the twin seeding pattern in
SimRadio::getRngSeed(). Applies to every sim instance uniformly (hero/B/R
all get proper per-session entropy now), not just the new relay path --
strictly an improvement with no compatibility concern, since identity
generation only ever runs once per fresh/empty IDBFS.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_018iubftDmKNWmkNnhJRz8UH

variants/sim/SimInstance.h

@@ -51,6 +51,35 @@ inline uint32_t sim_instance_salt() {
     return h >>> 0;
   });
 }
+
+// meshcore-solo-site's WebSocket relay bridge (Phase 3) exposed a real gap
+// in sim_instance_salt() alone: it's a pure function of the simInstanceTag
+// STRING ('hero'/'B'/'R'), so it's only ever useful for telling apart
+// same-tab instances that use different tags -- it's the SAME value every
+// time for two genuinely different browser tabs/machines that both boot a
+// 'hero' instance, which is exactly the new cross-visitor case. Combined
+// with SimRNG::begin()'s other two seed ingredients -- time(NULL) (1-second
+// resolution: two real visitors loading the page in the same second collide
+// outright) and `(uintptr_t)this` (a WASM linear-memory address, which is
+// fully deterministic across independent boots of the same binary doing the
+// same allocation sequence -- there's no ASLR inside a wasm sandbox, so this
+// contributes zero actual entropy, not "usually" different) -- two distinct
+// real visitors landing on the exact same wall-clock second reliably
+// produced byte-identical generated Ed25519 identities (confirmed while
+// testing the relay bridge: two independent, freshly-IDBFS browser contexts
+// launched together produced provably identical advert packets end to end).
+// A host page now passes one genuinely random value from the one place that
+// actually has real entropy per browser session -- crypto.getRandomValues()
+// -- as `simEntropy` on the Module config object (see meshcore-solo-site's
+// bootInstance()/bootRepeater()), read back here the same way
+// simInstanceTag already is. Defaults to 0 (this function's old, sole
+// behavior) if a host page doesn't set it, so nothing else changes.
+inline uint32_t sim_instance_entropy() {
+  return (uint32_t)EM_ASM_INT({
+    return (typeof Module !== 'undefined' && Module['simEntropy']) ? (Module['simEntropy'] >>> 0) : 0;
+  });
+}
 #else
 inline uint32_t sim_instance_salt() { return 0; }
+inline uint32_t sim_instance_entropy() { return 0; }
 #endif

variants/sim/SimRNG.h

@@ -10,12 +10,20 @@
 // in sim_instance_salt() so that two module instances of the SAME compiled
 // binary (the browser's two companion_radio instances) can't end up with
 // the same seed -- see SimInstance.h for why that's a real risk here, not
-// a hypothetical one.
+// a hypothetical one. sim_instance_entropy() (also SimInstance.h) mixes in
+// real crypto.getRandomValues()-sourced entropy from the host page on top
+// of that -- salt alone is a pure function of the tag string ('hero'/'B'/
+// 'R'), so it's identical across two genuinely different browser tabs that
+// both boot a 'hero' instance; combined with time(NULL)'s 1-second
+// resolution and a WASM heap pointer that's fully deterministic (no ASLR
+// inside the sandbox) across independent boots, two different real visitors
+// landing on the same wall-clock second used to get byte-identical
+// generated identities without this.
 class SimRNG : public mesh::RNG {
 public:
   SimRNG() { }
   void begin() {
-    unsigned seed = (unsigned)time(NULL) ^ (unsigned)(uintptr_t)this ^ sim_instance_salt();
+    unsigned seed = (unsigned)time(NULL) ^ (unsigned)(uintptr_t)this ^ sim_instance_salt() ^ sim_instance_entropy();
     srand(seed);
   }
   void random(uint8_t* dest, size_t sz) override {

variants/sim/SimRadio.h

@@ -116,7 +116,14 @@ public:
     // time(NULL) second, same `rand()` process state, often the same
     // `this` address across independent-but-identically-laid-out linear
     // memories) and end up with correlated "random" behaviour.
-    return (uint32_t)time(NULL) ^ (uint32_t)(uintptr_t)this ^ (uint32_t)rand() ^ sim_instance_salt();
+    // sim_instance_entropy(): same reasoning, but for two genuinely
+    // different browser tabs/machines both booting a same-tagged instance
+    // (e.g. two different visitors' 'hero') -- salt alone is a pure
+    // function of the tag string, identical for both; this mixes in real
+    // crypto.getRandomValues()-sourced entropy from the host page. See
+    // SimInstance.h and SimRNG.h for the full story (found while testing
+    // meshcore-solo-site's cross-visitor relay bridge).
+    return (uint32_t)time(NULL) ^ (uint32_t)(uintptr_t)this ^ (uint32_t)rand() ^ sim_instance_salt() ^ sim_instance_entropy();
   }
 
   void getFreqBounds(float& min_mhz, float& max_mhz) const {

variants/sim/build_wasm.sh

@@ -213,6 +213,11 @@ done
 # included in the build" -- confirmed by hitting exactly that during Phase
 # 3. Purely additive: nothing Phase 2's web/index.html already does
 # (sim_enqueue_key() with a plain number, no buffer marshaling) is affected.
+#
+# HEAPF32,HEAP32 (relay-bridge addition): sim_radio_get_params() writes its
+# freq/bw (float) and sf/cr (int) results into a JS-malloc'd scratch buffer
+# via out-params -- reading them back needs the typed-array views, same
+# reasoning as HEAPU8 above for sim_radio_poll_tx()'s byte buffer.
 "$EMXX" \
   "${OBJS[@]}" \
   -lidbfs.js \
@@ -222,7 +227,7 @@ done
   -sEXPORT_NAME=MeshCoreSim \
   -sENVIRONMENT=web \
   -sEXIT_RUNTIME=0 \
-  -sEXPORTED_RUNTIME_METHODS=FS,ccall,cwrap,HEAPU8 \
+  -sEXPORTED_RUNTIME_METHODS=FS,ccall,cwrap,HEAPU8,HEAPF32,HEAP32 \
   -sEXPORTED_FUNCTIONS=_main,_malloc,_free \
   -o "$OUT_DIR/meshcore_sim.js"