Frontend: Switch FP badge visibility to status-based gating

What does this MR do and why?

Frontend changes for Show FP badge for AI-detected flags regardless of confidence score, gated behind vulnerability_flag_status_based_gating feature flag.

Depends on !226166 (merged) (backend + feature flag introduction).

Key changes

Constants:

• Add FLAG_STATUSES, COMPLETED_DETECTION_STATUSES, PENDING_DETECTION_STATUSES, MANUAL_ORIGIN_PREFIX constants

GraphQL:

• Add status and origin fields to latestFlag fragment; remove unused description
• Add status field to dismiss false positive mutation response

Vulnerability detail page (header.vue):

• Switch detectedAsFalsePositive from confidence-score to status-based check (behind FF)
• Add hasActiveFpDetection computed to gate AI FP scan re-run buttons

Vulnerability report (vulnerability_list.vue):

• Switch hasAiExperimentSastFpDetection and shouldShowFalsePositiveBadge to status-based logic (behind FF)
• Use FLAG_STATUSES and MANUAL_ORIGIN_PREFIX constants

FP badge (ai_possible_fp_badge.vue):

• Add pending (info) and failed (danger) badge variants
• Hide confidence details for pending/failed states

FP details (false_positive_details.vue):

• Add pending and failed state rendering
• Gate dismiss button on completed detection status (behind FF)

Vulnerability details (vulnerability_details.vue):

• Switch shouldShowFalsePositiveDuoWorkflowDetails to status-based logic (behind FF)

How to set up and validate locally

1. Enable feature flags:

   Feature.enable(:ai_experiment_sast_fp_detection)
   Feature.enable(:vulnerability_flag_status_based_gating)

2. Create test data with various statuses:

   project = Project.find_by_full_path('YOUR_PROJECT_PATH')
   finding = project.vulnerabilities.first.finding
   
   # Pending scan
   finding.vulnerability_flags.create!(
     flag_type: :false_positive, origin: 'ai_sast_fp_detection',
     confidence_score: 0.0, status: :not_started,
     description: 'Scan not started', project_id: project.id
   )
   
   # Completed scan with 0.0 confidence (the bug this fixes)
   finding.vulnerability_flags.create!(
     flag_type: :false_positive, origin: 'ai_sast_fp_detection',
     confidence_score: 0.0, status: :detected_as_not_fp,
     description: 'Not a false positive', project_id: project.id
   )

3. Verify:

   • Pending flags show blue "FP Scanning" badge
   • Failed flags show red "FP Scan Failed" badge
   • Completed flags with confidence 0.0 show "Not an FP" badge (previously hidden)
   • Dismissed/manual-origin flags show no AI badge
   • Old behavior preserved when vulnerability_flag_status_based_gating is disabled

Screenshots / screen recordings

https://www.youtube.com/watch?v=0ZJsr00pLXs

MR acceptance checklist

• This MR does not add a migration
• Frontend changes have Jest spec coverage
• Changes are behind a feature flag
• GraphQL query complexity stays within budget