Add retry/abort measurement dialog for environment evaluation failures

[Copilot]

When a run hit transient failures while setting variables or reading observables, the subprocess surfaced a hard error and terminated the run. This change introduces an explicit operator decision point so failed measurements can be retried in-place instead of forcing immediate termination.

• Subprocess: retry-aware measurement loop

  • Added evaluate_measurement_with_retry(...) in core_subprocess.py.
  • Wrapped both initial-point and optimization-step evaluations with the retry loop.
  • On evaluation failure, subprocess publishes a structured measurement error event to the GUI queue and blocks for a control action (retry / abort).
  • If aborted, raises BadgerRunTerminated with failure context from the original measurement exception.

• GUI runner: operator action path

  • Extended BadgerRoutineSubprocess.check_queue() to detect structured measurement error messages.
  • Added action response messages back to the subprocess queue to continue (retry) or stop (abort) execution.
  • Preserved existing tuple-based error handling for non-measurement exceptions.

• New dialog: focused recovery UX

  • Added BadgerMeasurementRetryDialog with:
    • primary failure summary,
    • scrollable traceback/details,
    • explicit actions: Retry Measurement and Stop Run.
  • Dialog return value is mapped directly to subprocess control actions.

• Dialog tests

  • Added test_measurement_retry_dialog.py to cover retry/stop actions and verify rendered summary/details content.
  • Added a broader Badger application-flow test in test_gui_basic.py that verifies the retry dialog path is invoked from the routine runner queue handling and that the retry action is sent back to the subprocess channel.

# subprocess -> GUI
queue.put({
    "type": "measurement_error",
    "title": error_title,
    "traceback": error_traceback,
})

# GUI -> subprocess
self.data_and_error_queue.put({
    "type": "measurement_action",
    "action": "retry",  # or "abort"
})

[roussel-ryan]

@copilot add in a test to make sure the dialog works in the broader Badger application

[Copilot]

@copilot add in a test to make sure the dialog works in the broader Badger application

Added a broader Badger application-level test in test_gui_basic.py that exercises the measurement-error queue path through BadgerRoutineSubprocess.check_queue(), verifies the retry dialog is invoked, and confirms the retry action is sent back to the queue (37ef4dc).

Screenshot: