Watson Expansion and Borel Structure for Class A; Asymptotics of Class B in Non-Holomorphic Fractal Families

We derive the complete Watson expansion for the Class A stable-area law A(K) ~ C_A K-γ_A arising in non-holomorphic fractal families of Bird's classification. The expansion is a two-saddle decomposition: the interior saddle at t → 0 produces a convergent hypergeometric series G^A(w) = _2F_1(1/2, p; p + 3/2; w) in integer powers of w = K-(1+α)/2, whose Borel transform is entire; the endpoint boundary layer at t → 1 contributes a second convergent series in the incommensurable power w3/(1+α)^2, generically irrational and invisible to any finite truncation of the Watson series. The leading boundary-layer coefficient is C_bl = (2/3)√(2(1+α)) R3/2, proved in closed form. A complete set of Watson coefficients c_n = (1/2)_n (p)_n / ((p+3/2)_n n!) is derived via Beta recursion and confirmed to machine precision. For Class B we prove C_B = 2R exactly from the transcendental boundary equation and show the corrections form an iterated-logarithm series in ln(ln K)/ln K, qualitatively distinct from all other classes and implying an exotic Borel singularity structure. This is Paper 10 in a series on resurgent asymptotics for non-holomorphic fractal families. It closes Open Problems 1 and 2 of Paper 9 (doi:10.5281/zenodo.19222502). All computations performed on Chilly (NVIDIA RTX 3060 Ti, AMD Ryzen 5). Verified scripts: watson_correction_verify.py, watson_boundary_layer.py, borel_stokes_classA_v2.py, watson_pade_mpmath.py, bprime_stokes_verify.py, classB_En_extraction.py.v5 (erratum, 26 Mar 2026): d₁ = (2α−1)/4 in Theorem 3.1; was (α−1)/4 in v4. One-line algebra fix. No theorem, C_bl, incommensurability result, or numerical table affected except the d₁ column in §3.2. OP3 of §6.2 noted as closed in Paper 11.