feat: implement --dry-run mode and fix filename-to-stem mismatch in extract command

📋 Overview

This MR introduces a Dry Run mode across all primary upload commands and fixes a critical bug in the extracted text upload workflow that previously prevented correct record matching.

The update improves developer safety, usability, and reliability without affecting existing functionality.

🚀 Key Changes

🔹 CLI Enhancements

• Added --dry-run flag to:

  • upload-files
  • extract
  • resume

• CLI clearly indicates simulation mode:

  • Example: 📝 Extracted Text Upload (DRY RUN)

• Ensures users can validate workflows before execution

🔹 Simulation Logic

• Updated:

  • upload.py
  • extracted_text_upload.py

• Behavior in --dry-run mode:

  • Skips all API calls (POST, PUT)
  • Prevents any data mutation
  • Does NOT update local state files

• Added clear simulation logs:

[DRY RUN] Skipping upload for <filename>

• Ensures:

  • No side effects
  • Safe testing before real uploads

🐛 Bug Fix: Filename-to-Stem Mismatch

Issue: Extracted text upload failed to match CSV rows with JSON files due to mismatch between:

• Full filenames (with extensions)
• File stems (without extensions)

Fix:

• Standardized comparison using file stems
• Ensures correct mapping between CSV and JSON files

Impact:

• Fixes broken record matching
• Ensates successful extracted text uploads

🧪 How to Verify

✅ 1. Test Dry Run (Upload Files)

uv run corpus-client upload-files ./your-data --dry-run

• Should prompt for metadata
• Should NOT upload files
• Should display dry-run logs

✅ 2. Test Dry Run (Extract)

uv run corpus-client extract ./mapping.csv ./json_dir --dry-run

• Should process files
• Should NOT upload data
• Should show:

[DRY RUN] Skipping upload for <stem>

✅ Expected Output

╭────────────────────────────────────╮ │ 📝 Extracted Text Upload (DRY RUN) │ ╰────────────────────────────────────╯

🔍 Scanning 1 JSON files... 📤 Starting upload for 1 records... [DRY RUN] Skipping upload for transcription1 ✅ Extracted Text Upload Complete!

## ✅ Acceptance Criteria

* [x] `--dry-run` works across all commands
* [x] No API calls executed in dry run mode
* [x] Local state remains unchanged
* [x] Clear simulation logs displayed
* [x] Extract workflow correctly matches files
* [x] No regression in existing functionality


## 🎯 Result

* Safer bulk operations (pre-validation)
* Improved debugging and developer experience
* Fixed critical upload bug
* More reliable data processing pipeline