[NaNIsaac]

Tier: physics

• exp07's GenX script (runs/exp07_fourier_hires/engine/script_builder.py) now defines the background:
  • ibkg with getter and setter in Sim_Vars, starting from init_params["ibkg"];
  • the instrument starts from it instead of a fixed Ibkg=1e-10;
  • Sim() copies it into the instrument.
• Results change. engine/fitting.py registers "v.set_ibkg" as a fit parameter, and with the script on main GenX raises AttributeError: 'Sim_Vars' object has no attribute 'get_ibkg' when that parameter is registered (genx 3.7.13 as locked, and 3.8.0). The exp07 GenX fit therefore stopped in _setup_parameters. Now it can run, the background is fitted and applied, and the starting background is the 1e-7 the fitter passes rather than 1e-10.
• No exp07 run was redone (out of scope).

Fixes item 1 of #9. Item 2 (layer order) is a decision and is not touched.

Stacked on #6 (chore/conventions), which adds the tests directory and the pytest setup. Merge #6 first; GitHub then retargets this PR to main.

Correction to #9 item 1

The issue says the background "is fitted but not applied". That is the Reflecto case (Reflecto#2, fixed in Reflecto#5). In exp07 it is worse: the script never defined ibkg (git log -S ibkg on the file is empty), so the fit could not register the parameter at all.

Evidence

Test first (18f80c9), then the fix (0071c07). tests/test_exp07_genx_script.py:

test	main	this PR
test_background_can_be_registered_as_a_fit_parameter (what fitting.py does)	AttributeError	pass
test_background_reaches_the_simulation (ibkg 1e-7 → 1e-5 moves the q = 0.5 tail by 9.9e-6)	fails	pass
test_intensity_scale_reaches_the_simulation (control: i0 was already applied)	pass	pass

The tests simulate with compile=False. The default compile=True reruns the script and resets v, so it would hide any parameter change.

Tests

Local, Windows, uv sync --frozen (genx 3.7.13):

• uv run pytest: 27 passed (26 s);
• uvx ruff@0.17.1 check: clean.