GPTranslate: starejše različice za prenos

🚀 Major Azure OpenAI Support & Model Management Improvements

## ✨ New Features

### Enhanced Azure OpenAI Support
- **Full o1/o3 Reasoning Model Support**: Native support for Azure OpenAI o1, o1-mini, o3, and o3-mini models
- **Smart Parameter Detection**: Automatically uses max_completion_tokens instead of max_tokens for reasoning models
- **Developer Role Support**: Proper handling of system vs developer roles for different model types
- **Enhanced Endpoint Parsing**: Improved Azure endpoint URL parsing and validation
- **Auto-Model Addition**: Automatically extracts and adds models from pasted Azure endpoint URLs

### Modular Settings UI
- **Component-Based Architecture**: Refactored settings into modular Svelte components
- **Improved Model Management**: Dedicated model management interface with add/remove/toggle functionality
- **Real-Time Updates**: Model changes instantly reflect in the main window dropdown
- **Better Organization**: Tabbed settings interface for better user experience

### Live Model Management
- **Dynamic Model Updates**: Added/removed models appear immediately without app restart
- **Config State Management**: Centralized configuration management with reactive updates
- **Auto-Save**: Model changes are automatically saved and synchronized
- **No Default Models**: Clean slate approach - only shows user-configured models

## 🐛 Bug Fixes

### Azure OpenAI Fixes
- **Temperature Parameter**: Fixed temperature parameter errors for o1/o3 reasoning models
- **Token Limits**: Resolved max_tokens vs max_completion_tokens parameter conflicts
- **API Key Validation**: Improved Azure API key validation and error handling
- **Endpoint Compatibility**: Better support for both Cognitive Services and Models API endpoints

### UI/UX Improvements
- **Svelte Ownership Warnings**: Fixed prop mutation warnings in ModelSelector component
- **TypeScript Errors**: Resolved all TypeScript compilation issues
- **Model Refresh**: Fixed model dropdown not updating after settings changes

## 🔧 Technical Improvements

### Backend Enhancements
- **Conditional Parameters**: Smart parameter inclusion based on model type
- **Better Error Handling**: Enhanced error messages and logging for Azure configurations
- **Model Detection Logic**: Automatic reasoning model detection and parameter adjustment

### Frontend Architecture
- **Component Modularity**: Improved code organization with reusable components
- **State Management**: Better prop passing and state synchronization
- **Event-Driven Updates**: Reactive model management with immediate UI updates

## 📈 Performance & Reliability
- **Reduced Build Errors**: Eliminated TypeScript and compilation issues
- **Improved Validation**: Better input validation and error prevention
- **Enhanced Logging**: More detailed logging for debugging and troubleshooting

This release significantly improves Azure OpenAI compatibility, especially for the latest reasoning models (o1, o3 series), while providing a much better user experience for model management and configuration.

🐛 Bug Fixes:
• Fixed newline formatting issue with Ollama translations when typing in the "Original Text" field
• Improved Ollama response parsing to properly handle escaped newlines
• Enhanced text formatting consistency across all translation providers

🔧 Technical Improvements:
• Updated Ollama prompt instructions to preserve line breaks and formatting
• Added defensive programming for various newline escape formats (\n, /n, \r\n, \r)
• Improved translation response processing for better text formatting

This is a minor bugfix release that resolves the issue where Ollama would return literal "/n" instead of actual newlines when translating multi-paragraph text entered manually in the input field.

🚀 New Features:
• Ollama Support: Added complete support for Ollama local AI models for offline translation
• Provider Architecture: Refactored translation system with modular provider design
• Local Translation: Run translations completely offline with privacy-focused local models

🎯 Improvements:
• Provider UI: Beautiful new provider selection grid with professional icons
• Settings Redesign: Replaced dropdown with intuitive card-based provider selection
• Better Organization: Split translation code into separate provider modules
• Enhanced Config: Added ollama_url configuration with automatic defaults

🔧 Technical:
• Added ollama-rs crate integration for robust Ollama communication
• Improved error handling across all translation providers
• Enhanced validation system for all API endpoints including Ollama server connectivity

🐛 Bug Fixes:
• Azure API compatibility: Fixed "error sending request" issue by properly handling different Azure endpoint formats
• URL parsing: Robust parsing of Azure URLs with proper error handling
• Model flexibility: Removed restrictions that prevented using custom/newer models

🎯 Key Improvements:
• Better Azure Support: Now handles both Cognitive Services and Models API endpoints seamlessly
• User Experience: Real-time feedback when pasting Azure URLs with automatic extraction
• Visual Design: Professional header with larger SVG logo and better typography
• Flexibility: Removed hardcoded model restrictions, allowing any model name
• Error Handling: Comprehensive validation with clear error messages
• Icon update: Icon has been updated