Pattern reconciliation for normalizing duplicate identities.

This module provides operations to reconcile patterns by resolving duplicate identities and completing partial references. When patterns are parsed, streamed, or merged from multiple sources, the same identity may appear multiple times with different or evolving content. Reconciliation transforms such patterns into coherent ones where each identity appears exactly once.

Problem Statement

Currently nothing prevents the same identity from appearing multiple times with different content in a Pattern v. This can arise from:

• Parsing gram notation with duplicate node definitions
• Streaming pattern updates where identities evolve over time
• Merging patterns from different sources
• References: atomic patterns pointing to fuller definitions elsewhere

Solution Overview

The reconcile operation normalizes patterns using configurable policies:

1. **Collect**: Group all occurrences by identity
2. **Reconcile**: Resolve duplicates according to policy
3. **Rebuild**: Construct normalized pattern with each identity appearing once

Quick Start

>>> import Pattern.Core (Pattern(..), pattern, point)
>>> import Pattern.Reconcile
>>> import Subject.Core (Subject(..), Symbol(..))
>>> 
>>> -- Reconcile with LastWriteWins policy
>>> reconcile LastWriteWins patternWithDuplicates
>>> 
>>> -- Get detailed report
>>> let (result, report) = reconcileWithReport LastWriteWins pattern
>>> reportDuplicatesFound report  -- Number of duplicate identities

Reconciliation Policies

• LastWriteWins - Keep the last occurrence of each identity. Useful for streaming updates where the most recent value is authoritative.

• FirstWriteWins - Keep the first occurrence of each identity. Useful when initial definitions are authoritative.
• Merge elementStrategy valueStrategy - Combine all occurrences using configurable strategies for structure (elements) and content (value).
• Strict - Fail with detailed error if any duplicates have different content. Useful for validation and debugging.