Refactor replacements.json structure to eliminate duplication

[Helmi]

Description

The current replacements.json file has a confusing dual structure where project information is stored in both _auto_discovered.projects and project_names.replacements. This creates redundancy and makes the configuration harder to understand and maintain.

Context

• Project Area: Configuration system / Text-to-Speech replacements
• Complexity: 4/10 (Moderate - involves refactoring and migration)
• User Impact: Improves configuration clarity and maintainability

Current Problem

1. Projects are stored in two places:
   • _auto_discovered.projects (with rich metadata: folder, display_name, pronunciation)
   • project_names.replacements (simple key-value pairs)
2. The TTS system prioritizes _auto_discovered which is why user replacements work
3. Auto-discovery adds entries to BOTH locations, creating duplication
4. README documentation doesn't match the actual structure being used

Technical Details

Affected Files:

• src/ccnotify/notify.py - Core replacement logic (5 functions to update)
  • load_replacements() (line 430)
  • auto_add_project_to_replacements() (line 287)
  • apply_project_name_replacement() (line 445)
  • get_project_pronunciation() (line 457)
  • apply_command_replacement() (line 474)
• README.md - Documentation update for "Pronunciation Customization" section
• User config files - Will need automatic migration

Requirements

1. Consolidate project storage to a single projects section
2. Maintain rich metadata structure (folder, display_name, pronunciation)
3. Provide automatic migration for existing configurations
4. Add comprehensive examples for commands and patterns
5. Update documentation to reflect the new structure

Proposed Structure

{
  "projects": {
    "ccnotify": {
      "folder": "-Users-helmi-code-ccnotify",
      "display_name": "CCNotify",
      "pronunciation": "CC notify"
    },
    "agent-zero": {
      "folder": "-Users-helmi-code-agent-zero",
      "display_name": "Agent Zero",
      "pronunciation": "agent zero"
    }
  },
  "commands": {
    "ls": "list",
    "cd": "change directory",
    "rm": "remove",
    "mkdir": "make directory",
    "npm": "N P M",
    "uvx": "U V X"
  },
  "patterns": [
    {
      "pattern": "npm run (\\\\w+)",
      "replacement": "N P M run {1}"
    },
    {
      "pattern": "git (push|pull|commit)",
      "replacement": "git {1}"
    },
    {
      "pattern": "(.+)\\\\.py",
      "replacement": "{1} python file"
    }
  ]
}

Implementation Steps

• Update load_replacements() to use new default structure
• Modify auto_add_project_to_replacements() to write to projects section
• Update get_project_pronunciation() to read from projects[name].pronunciation
• Update apply_project_name_replacement() to use projects[name].display_name
• Add migration logic to convert old format to new on first load
• Add comprehensive command and pattern examples
• Update README documentation
• Test migration with existing configurations

Acceptance Criteria

• No duplicate project data in configuration
• Existing user configurations are automatically migrated
• TTS pronunciations continue to work correctly
• README accurately documents the new structure
• Rich metadata (folder, display_name, pronunciation) is preserved