Closes #264

This PR enhances the flexibility of column specification in pairtools by allowing CLI flags (-c1, -c2, etc.) to accept both column indices (ints) and column names (strs). Additionally, it refactors sorting and deduplication logic to use canonicalized column references, making pair_type (pt) optional for sorting.

🛠️ Changes Introduced

1. Modified CLI Flags (-c1, -c2, etc.)

   • Now accepts both integers and strings.
   • Default values remain integers for backward compatibility.

2. Implemented Conversion Functions in headerops

   • Converts between column indices and names.
   • Introduced a canonicalization function for consistent column handling.

3. Refactored sort and dedup Commands

   • Updated sorting and deduplication logic to work with both int- and str-based column references.
   • Ensured internal operations remain stable across different input types.

4. Made pair_type (pt) Optional for Sorting

   • Sorting no longer strictly depends on pair_type.
   • Verified correctness in cases where pair_type is missing.

5. Updated Tests

   • Added test cases for mixed column specification (ints & strs).
   • Verified correctness of sorting and deduplication without pair_type.

✅ How This Improves pairtools

• More user-friendly CLI: Users can now specify columns either by index or name, making scripts more readable.
• Backward compatibility maintained: Default values are still integers, ensuring existing workflows are not broken.
• Greater flexibility in data handling: Sorting and deduplication work smoothly even if pair_type is missing.

📌 Checklist Before Merge

• [] Code changes are complete and reviewed.
• [] New features are covered with unit tests.
• [] Backward compatibility is maintained.
• [] Documentation updates (if necessary).