fix: include excludes list in exported settings (fixes #2414)

[xovishnukosuri]

This PR fixes #2414 where the excludes list was not included when exporting profile settings.

Changes in src/vorta/profile_export.py:

• Export now includes ExclusionModel rows in the profile dict
• Import now restores exclusions after saving the profile
• Existing exclusions are cleared before import to avoid duplicates
• Made dict cleanup resilient via pop(..., None)

Added test in tests/unit/test_import_export.py:

• test_export_import_includes_exclusion_list verifies round-trip

Fixes #2414

[m3nu]

Good fix for #2414. The approach is clean and consistent with how SourceFileModel is already handled in the export/import flow. The del → pop(..., None) change also fixes a latent KeyError when importing with overwrite_settings=False from an export made with include_settings=False.

A few minor suggestions:

1. Backward compatibility when overwriting profiles

When importing an old export file (no ExclusionModel key) with overwrite_profile=True, existing exclusions on the target profile are silently preserved. This differs from SourceFileModel which always clears and replaces. Consider always clearing exclusions so that an overwrite is a full replacement:

ExclusionModel.delete().where(ExclusionModel.profile == self.id).execute()
if exclusions:
    for exclusion in exclusions:
        exclusion['profile'] = self.id
    ExclusionModel.insert_many(exclusions).execute()

2. Test could verify all round-tripped fields

The test only checks name. Asserting enabled and source too would give fuller confidence:

assert {(e.name, e.enabled, e.source) for e in imported_exclusions} == {
    ('*.tmp', True, 'custom'),
    ('Caches', True, 'preset'),
}

3. Nit: comment wording

# Restore source dirs replaces the old two-line comment that also mentioned deleting existing sources to avoid duplicates. The new comment is a bit less descriptive about the delete step.

[xovishnukosuri]

Thanks for the careful review. I’ve pushed updates: exclusions are now always cleared on overwrite for consistent behavior, the test asserts name+enabled+source, and I clarified the source-dir comment. Let me know if you want any tweaks.

[xovishnukosuri]

Pushed a small lint-friendly tweak: split the exclusion assertion into a named set variable. This should address the lint failure.