Bytecode Virtual Machine

# Bytecode Virtual Machine PRD

## Overview
Build a stack-based bytecode virtual machine with a complete instruction set, execution engine with call stack support, and a simple compiler that translates assembly-like text into executable bytecode.

## Requirements
- Define a bytecode instruction set: PUSH, POP, ADD, SUB, MUL, DIV, MOD, AND, OR, NOT, EQ, LT, GT, JMP, JMP_IF, CALL, RET, LOAD, STORE, PRINT, HALT
- Stack-based execution engine that processes instructions sequentially
- Call stack with stack frames supporting function calls and returns
- Local variables per stack frame with LOAD/STORE access
- Global variable storage accessible from any frame
- Simple assembly-like language parser that converts text source into bytecode
- Bytecode serialization — save compiled bytecode to binary file and load it back
- Program counter and main execution loop driving the VM
- Integer and floating-point arithmetic operations
- String operations including concatenation and length
- Comparison and boolean logic operations
- Error handling for stack overflow, invalid opcode, and division by zero
- Step-by-step debugger mode that prints stack state after each instruction
- Performance target: execute 1M+ instructions per second
- CLI runner that loads and executes bytecode or assembly files
- Disassembler that converts bytecode back to human-readable assembly

## Tech Stack
- TypeScript / Node.js
- No external libraries for the VM core — raw implementation required
- File system for bytecode persistence

## Scoring Criteria
- **Functional (40%)**: Instruction set executes correctly, function calls work, arithmetic is accurate
- **Quality (20%)**: Clean opcode design, robust error handling, well-structured code
- **Fidelity (25%)**: All instructions implemented, debugger works, serialization round-trips
- **Speed (15%)**: Time bonus