…s [IDE-1945]

Two defects stopped 'Reset overrides' from working:

1. No-op in sandboxed webview. VSCode renders the settings dialog in a
   sandboxed iframe without 'allow-modals', so window.confirm() is silently
   ignored and returns false. Both reset handlers gated their work behind
   confirm(), so the click bailed and nothing happened. Remove the confirm()
   prompts (reset is reversible and the dialog shows a disclaimer banner).

2. Incomplete reset. additional_parameters, additional_environment and
   scan_command_config are folder overrides editable in the dialog but were
   never cleared: the JS omitted them from FOLDER_RESET_FIELDS, and on the LS
   side they are processed by applyBasicFolderFields (marked handled), so the
   generic null-reset loop skipped them while their own handlers ignored a
   nil value -- making the reset unreachable. Add the three keys to
   FOLDER_RESET_FIELDS and teach applyStringField / applyStringSliceField /
   applyScanCommandConfig to Unset the override on {Changed:true, Value:nil}.

Extend the end-to-end reset test to cover the three non-scalar fields and
drop the dead win.confirm stubs / section-name formatter.

…derOverride

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

- docs: update reset field count 14→17, add additional_parameters,
  additional_environment, scan_command_config to the reset field list
- docs: correct incorrect claim that scan_command_config null-reset is
  a no-op; the LS handler now honours it via unsetFolderOverride
- tests: mirror FOLDER_RESET_FIELDS in RESET_FIELDS (14→17 entries)
- tests: remove AI-style bug annotation from test name
- tests: update "14 nulls" count to "17 nulls" in DOM-driven test name
- tests: rename test to say "auto-org" not "global org"

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Remove the duplicate "clicking reset triggers a save even with no other
edits" test (already covered by the DOM-driven no-edits test above it)
and add a test asserting the dirty tracker returns to clean after a
successful reset save.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

After applyPreferredOrg unsets preferred_org and org_set_by_user,
updateFolderConfigOrg was calling SetPreferredOrgAndOrgSetByUser("",false)
which re-added both keys as {Value:""/false, Changed:true} — making
HasUserOverride return true even though the user had just reset the
overrides.

Add a guard: only normalise to empty/false when at least one of the
keys is still present (i.e. a non-reset transition). If both are
already absent the call is a no-op for resolution purposes but creates
a spurious active-override entry in the store.

Also add SettingIsLspInitialized=true and DepScanStateAggregator to the
end-to-end reset test so it exercises the production code path through
updateFolderOrgIfNeeded.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

…figOrg

The else-if !currentSnap.OrgSetByUser branch was missing the same guard
added to the orgSetByUserJustChanged branch. After a reset clears both
overrides, a subsequent update changing only AutoDeterminedOrg would
trigger this branch and re-add preferred_org / org_set_by_user as
explicit {Value:""/false, Changed:true} entries, making HasUserOverride
return true again.

Apply the same guard: skip the normalisation call when both keys are
already absent. Add a regression test that seeds the exact scenario:
reset → AutoDeterminedOrg change → assert keys stay absent.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

VSCode sandboxed webviews block window.confirm() silently (no
allow-modals). Add ideBridge.confirm(message, callback) that routes
through the existing executeCommand bridge to snyk.showConfirmationDialog
in VSCode, falling back to window.confirm() in IDEs with no bridge
(JetBrains, Eclipse). Both reset handlers (section and folder override)
now gate their work behind this callback.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>