refactor: address the review of #147

[m-fehr]

Summary

Follow-up to #147 from its review.

• get_coordinates and get_nearest_index of Obstruction and SubObstruction take an optional quantity: node-based and cell-centred quantities have different coordinates. Without it they keep using the first quantity loaded, as before.
• Obstruction.vmin/vmax skip parts of the obstruction that have no data for the quantity, like get_global_boundary_data_arrays does, and raise KeyError only if no part has it.
• sim.meshes.get_global_boundary_data_arrays returns {} for a case without boundary data, like an obstruction without boundary data does. Mesh.has_boundary_data is new.
• Combining node-based and cell-centred patches of one quantity raises ValueError instead of placing them on one grid.
• The test inputs bndf_domain and bndf_global use DB instead of MB vents with several meshes (the FDS User Guide recommends DB for the domain boundary) and one namelist per line, rerun with FDS 6.11.1.

Tests

New tests for the coordinates of a chosen quantity, the orientation filter of Obstruction.get_boundary_data, parts without the quantity, mixed node-based and cell-centred patches, and a case without boundary data.

🤖 Generated with Claude Code

[codecov]

Codecov Report

❌ Patch coverage is 89.65517% with 3 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
fdsreader/bndf/obstruction.py	87.50%	3 Missing ⚠️

📢 Thoughts on this report? Let us know!