♻️ Consolidate OpenQASM import around the compiler frontend

[burgholzer]

🤖 AI text below 🤖

Goal

Reduce the duplicated OpenQASM parser and semantic implementation while preserving lightweight QuantumComputation consumers.

Direction to evaluate

• Make the compiler's OpenQASM parser and semantic analysis the canonical frontend.
• Provide separate emitters for MLIR programs and QuantumComputation rather than requiring every CoreIR consumer to link the complete LLVM/MLIR stack.
• Retire the legacy parser only after behavioral parity is demonstrated.
• Coordinate with the compiler translation work tracked in ✨ Extend OpenQASM-to-QC translation #1733.

Required compatibility

• OpenQASM 2 and supported OpenQASM 3 input.
• Python QuantumComputation imports.
• DDSIM inputs and layout comments.
• Symbolic parameters, custom gates, control flow, diagnostics, and source locations to the extent supported today.

Acceptance criteria

• One lexer/parser/semantic implementation serves both compiler and classic IR paths.
• Lightweight CoreIR consumers do not acquire a runtime LLVM/MLIR dependency merely to parse OpenQASM.
• Differential tests cover accepted programs, rejected programs, diagnostics, and constructed circuit semantics before the old frontend is deleted.