Add bulk import for expert opinions (CSV/JSON)

[caitlon]

User Story

As a project admin, I want to import multiple expert opinions at once via CSV or JSON file, so that I can efficiently set up projects with many experts without manual entry.

Problem Statement

Currently, admins must add expert opinions one by one through the UI. For projects with 20+ experts (like the budget and pendlers case studies), this manual process is time-consuming and error-prone.

Proposed Solution

Data Format Support

CSV Format:

expert_name,lower,peak,upper,comment
Expert1,5.0,10.0,15.0,Optional comment
Expert2,8.0,12.0,18.0,

JSON Format:

{
  "opinions": [
    {"name": "Expert1", "lower": 5.0, "peak": 10.0, "upper": 15.0, "comment": "Optional"},
    {"name": "Expert2", "lower": 8.0, "peak": 12.0, "upper": 18.0}
  ]
}

Backend Requirements

• POST /api/v1/projects/{project_id}/opinions/bulk
• Validate all opinions before inserting (atomic transaction)
• Return detailed error report with row numbers for failed validations
• Enforce fuzzy number constraint: lower <= peak <= upper
• Respect project's scale_min and scale_max settings

Frontend Requirements

• Drag-and-drop file upload component
• Preview table showing parsed data before import
• Progress indicator during import
• Error summary with clickable row navigation
• Download template button (CSV and JSON examples)

Acceptance Criteria

• User can upload CSV file with expert opinions
• User can upload JSON file with expert opinions
• Preview shows all rows with validation status (valid/invalid)
• Invalid rows are highlighted with error messages
• User can fix invalid rows inline before submitting
• Successful import adds all opinions atomically (all or nothing)
• Failed import shows which rows failed and why
• UI supports both English and Czech languages
• API endpoint has 100% test coverage
• API documentation updated in api/README.md

Definition of Done

• PR passes all CI checks (lint, types, tests)
• CodeRabbit review addressed
• API tests cover success, validation errors, partial failures
• Frontend tested manually on Chrome and Firefox
• i18n translations added for both EN and CS

Technical Notes

• Use existing OpinionService patterns from api/services/opinion_service.py
• Follow validation patterns from api/schemas/validators.py
• Frontend file upload can use native <input type="file"> with drag-drop wrapper

Related

• Case study data format: examples/data/README.md
• Opinion schema: api/schemas/opinion.py