Add GUI control panel with live measurements and full instrument control

New `mppt-gui` command launches a tkinter app with:
- Live numeric readouts for all 13 measurement channels
- 4-panel real-time matplotlib graphs (power, efficiency, voltage, current)
- Full supply control: voltage, current, OVP, slew, output on/off
- Full load control: mode (CC/CR/CV/CP), setpoint, slew, load on/off
- Meter settings: wiring mode, response speed, coupling per channel
- CSV data logging with start/stop control
- Red SAFE OFF button (also Escape key) bypasses command queue
- Threaded architecture: instrument I/O never blocks the GUI
- Fix apply() silently failing: use set_current + set_voltage instead

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

testbench/bench.py

@@ -384,7 +384,8 @@ class MPPTTestbench:
         elif i_start > i_stop and i_step > 0:
             i_step = -i_step
 
-        self.supply.apply(voltage, current_limit)
+        self.supply.set_current(current_limit)
+        self.supply.set_voltage(voltage)
         self.supply.output_on()
         self.load.set_mode("CC")
         self.load.set_cc_current(i_start)
@@ -439,7 +440,8 @@ class MPPTTestbench:
             Dict with avg_input_power, avg_output_power, avg_efficiency,
             plus individual supply/load readings.
         """
-        self.supply.apply(voltage, current_limit)
+        self.supply.set_current(current_limit)
+        self.supply.set_voltage(voltage)
         self.supply.output_on()
         time.sleep(settle_time)