feat: implement configuration-driven compression thresholds

[ooples]

Summary

Implement configuration-driven thresholds for token optimization operations, following the pattern discovered in Gemini CLI.

Background

Currently, hardcoded values are scattered throughout the codebase. Gemini CLI uses a centralized configuration system with schema validation.

Gemini CLI Pattern Reference

File: packages/core/src/config/settingsSchema.ts

• Uses JSON schema for configuration validation
• Defines CompressionTokenThreshold (default: 0.7 = 70%)
• Defines CompressionPreserveThreshold (default: 0.3 = keep 30%)
• Defines per-model tokenBudget limits

Implementation

Create ~/.token-optimizer/config.json:

```json
{
"compressionTokenThreshold": 0.7,
"compressionPreserveThreshold": 0.3,
"minTokensBeforeCompression": 1000,
"modelTokenLimits": {
"gemini-2.5-flash": 1000000,
"gemini-1.5-pro": 2000000
},
"cacheSettings": {
"maxSize": 1000,
"ttlSeconds": 3600
},
"optimizationSettings": {
"minOutputSizeBytes": 500,
"quality": "balanced"
}
}
```

Create PowerShell config class in src/Config.ps1.

Acceptance Criteria

• Configuration schema with validation
• PowerShell Config class with Load/Save
• Default config created on first run
• Orchestrator uses config values
• Config updatable without code changes
• Unit tests for config validation

Priority

High - Foundation for other optimizations

References

Gemini CLI: packages/core/src/config/settingsSchema.ts