v0.2: Refactor into class-based

[janitha-mahanthe]

Summary

Refactor AutoREACTER from script-based workflow into a modular, class-based architecture.

This refactor will:

• Separate core components (InputParser, Detector, ReactionSelection, TemplateBuilder, LUNARClient).
• Remove hidden state mutations.
• Enable visualization layers (v0.2: Notebook visualizations for reactant/product templates #24, Add Functional Group Reaction Visualization Before Selection #34).
• Fix workflow ordering issues (Refactor Detector Execution After Input Validation + Restore Reaction Selection Prompt #31, Preserve Original Reaction Indices When User Drops Reactions #33).
• Provide a clean foundation for future force-field flexibility.

All major workflow stages must become independent, testable classes with clear interfaces.

---

Blocking

This refactor is required before:

• v0.2: Update atom typing to modern types #26 (Atom typing compatibility update)

---

Milestone

v0.2-beta