docs(ctp): CTP1-CTP8 checked against the published paper and Deb's 2001 book

[tachsin]

CTP8 came from the authors' code only, and CTP1-CTP7 were checked against the preprint (KanGAL report 200005). They are now checked against Deb, Pratap and Meyarivan's published paper (EMO 2001, LNCS 1993, pp. 289-298) and Deb's book (Multi-Objective Optimization Using Evolutionary Algorithms, Wiley 2001, section 8.3.5, pp. 352-360). This closes the open item in #260.

All eight definitions match both sources. No code, optimal_front or example output changed.

• CTP1 and the CTP2-CTP7 generator: the same equations as the preprint. CTP1 is the paper's eq. 4 and the book's eq. 8.45; the generator is eq. 5 and eq. 8.46.
• CTP8 (book p. 358): C₁: θ = 0.1π, a = 40, b = 0.5, c = 1, d = 2, e = −2. C₂: θ = −0.05π, a = 40, b = 2, c = 1, d = 6, e = 0. These are exactly the values taken from the code. The three front pieces marked in the book's figure 232 match genoxide's derived pieces.

Documented, not changed:

• f₂: both sources print f₂ = g(1 − f₁/g), while their figures draw 1 − √f₁, as the code does. genoxide keeps the square root.
• g and the bounds: neither source fixes them; genoxide keeps the code's.
• Figures: they omit CTP5's last touch point and CTP7's edge pieces, which genoxide's fronts include.

Fixed: CTP8's documented piece ends were slightly off (0.1341 for 0.1345, and similar). The docs, the Python docstrings, the eight examples' READMEs and comments, and the plan now cite the paper, the report and the book with pages. CTP8 is marked as verified against its original.