Add onboarding instructions for Copilot coding agents

[Copilot]

Adds .github/copilot-instructions.md to provide coding agents with essential context for effective contribution. Addresses the repository's domain complexity (gamma-ray astronomy + ML) and critical implementation details (dual telescope indexing modes).

Structure

• Domain-specific sections for three reviewer personas:

  • Gamma-ray astronomy: air shower physics, VERITAS/CTAO telescope arrays, stereoscopic reconstruction
  • Machine learning: XGBoost regression/classification, energy-binned training, telescope multiplicity handling
  • Python/computing: ROOT→awkward→pandas pipeline, parallel decompression, variable-length array handling

• Critical gotchas: Telescope indexing (VERITAS fixed vs CTAO variable), DispTelList_T mapping, NaN handling, feature sorting requirements

• Verified workflows: Build/test/lint commands, CLI entry points, changelog management (towncrier), pre-commit hooks

Key Technical Details Documented

Telescope indexing (most complex logic):

• VERITAS mode: Fixed indices with R_core branch
• CTAO mode: Variable-length with ImgSel_list branch
• _resolve_branch_aliases() and _normalize_telescope_variable_to_tel_id_space() handle compatibility

Data pipeline:

# Variable-length telescope data → dense arrays
flatten_feature_data()  # Sorts by mirror area, then size
# Missing telescopes padded with np.nan (XGBoost native support)

Test coverage: 27% overall (60% for core data_processing.py, 0% for config.py/models.py/evaluate.py)

Files Changed

• .github/copilot-instructions.md: 492 lines covering project overview, conventions, domain knowledge, common patterns
• .gitignore: Added .coverage test artifact

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.