Integrate export size and token estimation into the selection workflow

Release v0.1.6 introduces total export size and token count estimation models to help users gauge codebase sizes relative to LLM context window limits before initiating exports.

Key Features & Enhancements:

1. Token & Size Estimation Engine (apps/desktop/src/renderer/lib/tokenEstimate.ts)
   • Implemented a dependency-free token estimation utility using the standard heuristic of 1 token per 4 bytes.
   • Defined precise export wrapper overhead constants, including a 500-byte header overhead and 120-byte per-file Markdown formatting overhead.
   • Added state classifiers to calculate context status against common LLM context window thresholds: 128K, 200K, and 1M tokens.
   • Added utility functions to format raw bytes and token counts into human-readable strings.

2. Selection Summary Calculation (apps/desktop/src/renderer/lib/selection.ts)
   • Extended the selection model summary to aggregate raw size measurements of all checked file nodes.
   • Automatically integrated export header and markdown wrapper overhead calculations into the total byte counts.
   • Integrated the estimation engine to expose estimated total bytes and token counts dynamically.

3. Ul-level Statistics & Context Badges (apps/desktop/src/renderer/App.tsx)
   • Added user interface rows for rendering human-readable totals for export size and estimated tokens.
   • Rendered dynamic, color-coded badges (green, amber, or red status flags) corresponding to 128K, 200K, and 1M token limits.
   • Added styles for the statistics row, alignment details, and limits state alerts.

4. Testing & Verification (apps/desktop/tests/tokenEstimate.test.ts, apps/desktop/tests/selection.test.ts)
   • Implemented a dedicated suite of unit tests checking boundary conditions, ceiling rules, formatting, and badge transitions.
   • Updated the selection model tests to verify correct total byte accumulation and metadata overhead insertion.