Live capture on real DSO-2250 hardware reports no waveform

[malikalhack]

• Symptom: GetCaptureState polling never leaves the empty-buffer state on a
  physical Hantek DSO-2250 (state byte stays 0x00), so no channel data is
  ever read and the display keeps showing no waveform, while Demo mode and
  every deterministic CTest continue to pass.
• Investigated and ruled out across several verified, warning-free
  Debug/Release builds, each checked against fresh USB captures from the
  physical device: a missing ForceTrigger before polling, an incomplete
  configureCapture() command sequence, missing empty-response retry
  handling, resending TriggerEnabled on every poll, and a missing
  SetOffset (CONTROL_SETOFFSET, request 0xB4) channel/trigger offset
  command derived from the device's calibration table - all implemented,
  hardware-tested, and individually confirmed insufficient.
• Found by correlating each USB command byte with its immediately following
  response byte across a full Windows reference driver capture (naive
  per-endpoint response counts are misleading, because GetCaptureState,
  GetChannelData, and the unrelated logic-channel/auto-range subsystem all
  share bulk-IN endpoint 0x86): the reference driver's completed-capture
  state value is 3, not the previously assumed 2. Corrected
  captureCompleteState from 2U to 3U for both DSO-2250 profiles in
  usb/src/usb_device.cpp.
• This correction is warning-free in Debug/Release and passes all CTest
  cases, but still did NOT resolve the symptom on hardware retest - the
  root cause remains open. Treated as a known, unresolved bug pending
  further hardware-side USB capture analysis.

Wireshark-dumps.zip