fix(sim): wire ATtiny85 PBx pin edges into the SPICE mixed-mode loop

arduinoPinToName() had no ATtiny85 case, so pin 1 reverse-mapped to "1"
instead of "PB1" (the wire/netlist name). The MCU-edge listener was never
attached (name not in pinsInCircuit) and the SPICE V-source was never altered
on digitalWrite, so a blink LED's branch current stayed at its HIGH value —
the LED latched ON and never turned off (and analogWrite duty changes never
re-solved). Map attiny85 pin N -> "PBN", mirroring pinNameToArduinoPin.

frontend/src/simulation/spice/connectMcuEdgesToService.ts

@@ -84,6 +84,15 @@ export function connectMcuEdgesToService(service: CircuitSimulationService): ()
     if (boardKind.startsWith('esp32')) {
       return `GPIO${arduinoPin}`;
     }
+    // ATtiny85 wires reference port-style names (PB0..PB5), matching the
+    // netlist pin names from collectPinStates. Without this, the reverse
+    // mapping returns "1" instead of "PB1", so the MCU-edge listener is
+    // never attached (pin name not in `pinsInCircuit`) and the SPICE
+    // V-source is never altered on digitalWrite LOW — the LED latches ON
+    // (and analogWrite duty changes never re-solve). See pinNameToArduinoPin.
+    if (boardKind === 'attiny85') {
+      return `PB${arduinoPin}`;
+    }
     return String(arduinoPin);
   }