Fix pcap timestamp wrap, ATR-hunt EOF crash, start.bat protocol; tighten launchers + vs_reader

- pd.py ts_from_samplenum: drop c_uint32 truncation + dead sampleOverflowCount
  (pcap timestamps wrapped backwards past ~4.5 min at 16 MHz).
- pd.py handle_atr: guard None reads at EOF (was TypeError (None & 0x80) on a
  capture truncated mid-ATR-hunt).
- start.bat: add --protocol (+ :protocol=T=0 default) to match start.sh's
  mid-session decode; correct the 'flag-for-flag' header comment.
- start.sh: reject --rst/--vcc assigned to the same pin (was unchecked).
- setup.bat: set DEVICE_MSG on blocked/skipped probe so the trailing
  'unknown' message is suppressed.
- vs_reader.py: RESULT now fails on CONCAT / payload mismatch / suspicious
  CLA, not just GARBAGE (OK text unchanged; clean traces stay identical).

setup.bat

@@ -106,10 +106,12 @@ for %%r in (1M 2M 4M 8M 12M 16M 20M 24M) do (
         -P "iso7816:clk=D0:data=D1" >"%TEMP%\iso7816-probe.txt" 2>&1
     type "%TEMP%\iso7816-probe.txt" | findstr /c:"Unable to claim" >nul && (
         echo blocked - another program holds the device
+        set "DEVICE_MSG=1"
         goto :probedone
     )
     type "%TEMP%\iso7816-probe.txt" | findstr /c:"Failed to open" /c:"No devices found" >nul && (
         echo skipped - no FX2 device connected
+        set "DEVICE_MSG=1"
         goto :probedone
     )
     type "%TEMP%\iso7816-probe.txt" | findstr /r /c:"invalid argument" /c:"Unable to sample" /c:"Could not start" >nul || set MAX_RATE=%%r