=== 1. report example p. 1457 (transcribed by the auditor from 10x renders) ===
     .###b...
     ##b##...
     ###r####
     r#######
     #######b
     ######b#
     .r####..
     ...##b..
   side 8 | digitally convex (footnote 2): True
   sigma in S_lambda: True
   A-pattern occurrences in lambda: [] -> sigma in A_lambda: True (caption: yes)
   (in the full square sigma has 4 A-occurrences, i.e. contains 123)
   red / blue occurrences: [] [] -> coloured sigma in B_lambda: True (caption: yes)
   occurrences of the 4-point pi: [(1, 3, 7, 8), (1, 4, 6, 7), (1, 4, 6, 8), (1, 4, 7, 8), (1, 5, 6, 7), (1, 5, 6, 8), (1, 5, 7, 8), (4, 5, 6, 7), (4, 5, 6, 8), (4, 5, 7, 8)] (figure circles columns 1,4,6,7)
   crosses of that occurrence land on [(7, 4), (4, 8)] (figure draws crosses at (7,4) and (4,8))

=== 2. remark p. 1456: 4x4 square minus bottom-left staircase {(1,3),(1,4),(2,4)}, sigma = identity ===
   identity in A_lambda: True
   valid 2-colourings of the identity: [] (report: none)
   every colouring with two equal adjacent colours is invalid ("must alternate"): True

=== 3. the claimed counterexample: 7x7 band -3 <= y - x <= 4 (y = row from top) ===
     ####...
     #####..
     ######.
     #######
     #######
     .######
     ..#####
   cells 40 | side 7 | digitally convex (footnote 2): True
   |S_lambda| = 1066   A = 536   B = 515   (#sigma with >=1 good colouring = 186)   [0.1s]
   B < A : True

=== 4. brute-force census of ALL subsets of [n]^2, n <= 4, with the footnote-2 test ===
   side 1: digitally convex subsets of [m]^2 with side m: 1 (with full 1x1 bounding box: 1); min(B-A) = 1 (A=1, B=2)
   side 2: digitally convex subsets of [m]^2 with side m: 9 (with full 2x2 bounding box: 5); min(B-A) = 0 (A=0, B=0)
   side 3: digitally convex subsets of [m]^2 with side m: 126 (with full 3x3 bounding box: 68); min(B-A) = 0 (A=0, B=0)
   side 4: digitally convex subsets of [m]^2 with side m: 2156 (with full 4x4 bounding box: 1110); min(B-A) = 0 (A=0, B=0)
[total 0.5s]
