mctest: say that -n on a %Test/%Example line is ignored

[willend]

Summary

mctest only uses the -n on %Scan/%TestScan lines. A %Test/%Example always runs at the global --ncount (scaled by --statfactor). This was never stated anywhere, and about 40 shipped %Example lines carry a -n that has never been used (1e8 in FRMII_SPODI, 1e7/1e5/1e4/1e3 elsewhere, in both McStas and McXtrace). This came up while fixing templateSANS_Mantid's reference for #2808.

History: the -n has always been ignored. Before split_test_line (e5e7b58), the line's -n was passed to mcrun ahead of the global -n, and mcrun (optparse) keeps the last value given, so the global one won.

Changes

• mctest.py: log one line for each %Test/%Example whose -n is ignored, and add a comment at the ncount selection.
• mctest.md: say that -n on a %Test: line is only a hint for users (mcdoc shows it) and is ignored, and that better test statistics come from changing the instrument (e.g. SPLIT).

No change to which ncount tests run at or to their results. The -n tokens in the instruments are left as they are, since they serve as advice for users.

Honouring -n instead was considered and rejected. It would run FRMII_SPODI 100 times longer (very likely a timeout), run several instruments 10 times longer, and run the 1e3–1e5 tests with 10 to 1000 times fewer rays than their 1e6 targets were made with.

Testing

• py_compile passes.
• Ran test parsing on FRMII_SPODI, Samples_Incoherent, Test_Magnetic_Majorana, SE_example (McStas) and Test_Absorption (McXtrace). The message is logged for the %Example lines with -n, and not for scans.
• Not run: a full mctest/mxtest run.

🤖 Generated with Claude Code