Run the corpus sweep through the two-route conversion

[peterbjohnson]

Every ticket writes its tests before its code, and its pull request reports a run over a target set or a corpus folder with the number of flagged fields. See docs/plan.md.

The corpus command runs the folder conversion of ticket 1 over each set in the corpus and
writes results.csv with one row per sheet and these columns: set, sheet, questions, parts,
fields, agreed, adjudicated, flagged, not verbatim, tokens, seconds, and a reason where the
sheet produced no set. The ticket is done when the three ExampleContents folders have been
swept and the pull request shows the table.

---

Workbench ticket t39.

Live sweep over the three ExampleContents folders (manager, 2026-09-23 13:05)

19 sheets (solutions files paired), every one built a set, exit 0. Totals: 1006 fields, 477 agreed, 118 adjudicated, 105 flagged, 12 not verbatim, 830,728 tokens.

built    MECH60014_Stress_analysis_3/Stress_Sheet1_Sol_UH.pdf: 59 fields, flagged 0  route B failed: Unknown input format pdf Pandoc can convert to PDF, but not from PDF.
built    MECH60014_Stress_analysis_3/Stress_Sheet1_UH(1).pdf: 59 fields, flagged 0  route B failed: Unknown input format pdf Pandoc can convert to PDF, but not from PDF.
built    MECH60014_Stress_analysis_3/Stress_Sheet3 (CB).pdf: 43 fields, flagged 0  route B failed: Unknown input format pdf Pandoc can convert to PDF, but not from PDF.
built    MECH60014_Stress_analysis_3/Stress_Sheet3_Sol(1).docx: 46 fields, flagged 12
built    MECH60014_Stress_analysis_3/Stress_Sheet4.docx: 15 fields, flagged 5
built    MECH60014_Stress_analysis_3/Stress_Sheet5 (2).docx: 50 fields, flagged 22
built    MECH60014_Stress_analysis_3/Stress_Sheet6 (1).docx: 41 fields, flagged 7
built    MECH60014_Stress_analysis_3/Stress_Sheet6_ Sol (1).docx: 47 fields, flagged 7
built    PHYS40002-Mechanics/problem_sheets_and_figures/mechanics_23-24_PS1.tex: 65 fields, flagged 3
built    PHYS40002-Mechanics/problem_sheets_and_figures/mechanics_23-24_PS2.tex: 103 fields, flagged 12
built    PHYS40002-Mechanics/problem_sheets_and_figures/mechanics_23-24_PS3.tex: 53 fields, flagged 4
built    PHYS40002-Mechanics/problem_sheets_and_figures/mechanics_23-24_PS4.tex: 96 fields, flagged 9
built    PHYS40002-Mechanics/problem_sheets_and_figures/mechanics_23-24_PS5.tex: 37 fields, flagged 3
built    PHYS40002-Mechanics/problem_sheets_and_figures/mechanics_23-24_PS6.tex: 53 fields, flagged 1
built    PHYS40002-Mechanics/problem_sheets_and_figures/mechanics_23-24_PS7.tex: 46 fields, flagged 5
built    PHYS40002-Mechanics/problem_sheets_and_figures/mechanics_23-24_PS8.tex: 75 fields, flagged 13
built    PHYS40002-Mechanics/problem_sheets_and_figures/mechanics_23-24_PS9.tex: 15 fields, flagged 1
built    UCL_MechEng/Worksheet_1.pdf: 63 fields, flagged 1  no filter: pandoc reads no sheet of this set
built    UCL_MechEng/Worksheet_2.pdf: 40 fields, flagged 0  no filter: pandoc reads no sheet of this set
19 sheets, written to /private/tmp/claude-501/-Users-peterbjohnson-code-lambdafeedback-in2lambda/94029b0b-e025-469e-ac13-85e39254e374/scratchpad/t39-results.csv
exit 0

Route B did not run on the PDF sets: the filter is applied to the PDF itself rather than to its OCR markdown ('Unknown input format pdf', 'pandoc reads no sheet of this set'). Recorded as a draft ticket.