Add detection ROIs to existing ROIs instead of replacing them

[StefanoPierini-Codra]

Summary

When creating regions of interest (ROIs) from contour detection or blob
detection, DataLab currently replaces any pre-existing ROIs on the image.
If ROIs are already defined, a confirmation popup warns the user that the
existing ROIs will be lost.

I propose to change this behavior so that newly detected ROIs are appended
to the existing ROIs instead of replacing them. As a consequence, the
confirmation popup is no longer needed and can be removed.

Current behavior

• Running a detection with ROI creation enabled (create_rois=True) on an image
  that already has ROIs triggers a warning dialog:

  Regions of interest are already defined for this image.
  Creating new ROIs from detection will replace the existing ones, which will
  be lost.
  Do you want to continue?

• If the user confirms, the existing ROIs are discarded and replaced by the
  newly detected ones.

Proposed behavior

• Newly detected ROIs (from contour and blob detection) are added to the
  image's existing ROIs rather than overwriting them.
• The confirmation popup in preprocess_1_to_0 becomes unnecessary and is
  removed.

Rationale

• Appending is non-destructive: the user never loses previously defined ROIs.
• Removing the popup streamlines the workflow (one less confirmation click).

Affected repositories

• DataLab (DataLab-Platform/DataLab): GUI confirmation dialog and tests
• Sigima (DataLab-Platform/Sigima): apply_detection_rois ROI merge logic

[StefanoPierini-Codra]

Summary

With Sigima now appending detected ROIs to existing ones (non-destructive), the DataLab confirmation dialog that warned about ROI replacement is no longer needed. This PR removes it and refocuses the related tests on the append behavior.

Changes

• datalab/gui/processor/image.py: removed the ImageProcessor.preprocess_1_to_0 override (the "Regions of interest are already defined… will replace the existing ones" confirmation dialog). The inherited BaseProcessor.preprocess_1_to_0 hook (returns True) is used, so detection proceeds without prompting. Removed two now-unused imports (guidata.dataset, datalab.env).
• Tests (detection_roi_replace_unit_test.py → detection_roi_merge_unit_test.py): dropped the 3 dialog-centric tests, rewrote the "existing ROI" case as test_create_rois_appends_to_existing_roi (existing ROIs preserved + count grows), kept the no-ROI / create_rois=False / auto-recompute / contour / no infinite loop tests.
• datalab/locale/fr/LC_MESSAGES/datalab.po: removed only the now-obsolete ROI-replacement dialog string.

Notes

• auto_recompute_analysis still forces create_rois=False (prevents ROI accumulation on silent recomputes).
• Depends on Sigima "Delete all" action enable state is sometimes not refreshed #36.