OpenAI GPT-5 improvements (#9)

A series of fixes for making GPT-5 reliable and effective with anyclaude.

## UX changes

- Converts opaque OpenAI service errors into 429s so CC natively retries instead of failing. These are very common.
- Rich debugging support via `ANYCLAUDE_DEBUG`
- Supports specifying reasoning effort and service tier 

## Codebase

- Added CI

## Remaining issues

- GPT-5 often fails to use the native tool calls. These failures seems intermittent.

CLAUDE.md

@@ -34,6 +34,9 @@ bun install
 # Build the project (creates dist/main.js with shebang)
 bun run build
 
+# Run the built binary
+bun run ./dist/main.js
+
 # The build command:
 # 1. Compiles TypeScript to CommonJS for Node.js compatibility
 # 2. Adds Node shebang for CLI execution
@@ -64,3 +67,6 @@ Required for each provider:
 Special modes:
 
 - `PROXY_ONLY=true`: Run proxy server without spawning Claude Code
+- `ANYCLAUDE_DEBUG=1|2`: Enable debug logging (1=basic, 2=verbose)
+
+- OpenAI's gpt-5 was released in August 2025