fix(radiative): only apply endpoint resummation within delta_cut per spec Section 3.4

The resummation was previously applied to all energy points when the
flag was set, causing resummed and standard modes to differ everywhere.
Per spec, it should only replace C(beta)*ln(W0-W) with the resummed
power-law form when (W0-W) < delta_cut (~1e-3).

Adds delta_cut parameter (default 1e-3) and corrects the resummation
switch test to verify identical results far from endpoint.

beta_spectrum/components/radiative.py

@@ -33,6 +33,7 @@ class RadiativeCorrection(SpectrumComponent):
         Z: int = 0,
         A: int | None = None,
         use_endpoint_resummation: bool = True,
+        delta_cut: float = 1e-3,
     ):
         """
         Args:
@@ -40,11 +41,13 @@ class RadiativeCorrection(SpectrumComponent):
             Z: Nuclear charge of daughter nucleus
             A: Mass number (optional, affects nuclear model correction)
             use_endpoint_resummation: Apply Eq.(53) to handle ln(W0 - W) divergence
+            delta_cut: Only apply resummation when (W0 - W) < delta_cut (spec Section 3.4)
         """
         self.W0 = W0
         self.Z = Z
         self.A = A
         self.use_endpoint_resummation = use_endpoint_resummation
+        self.delta_cut = delta_cut
         self.m_p = MP_MEV / ME_MEV  # ~1836.15 (proton mass in m_e units)
         self.gamma_E = 0.57721566490153286  # Euler-Mascheroni constant
 
@@ -119,14 +122,20 @@ class RadiativeCorrection(SpectrumComponent):
         if self.use_endpoint_resummation:
             # Apply soft-photon resummation from Sirlin (1987)
             # Replace C(beta)*ln(W0-W) with C(beta)*((W0-W)^{t(beta)}-1)/t(beta)
+            # Only near endpoint: (W0 - W) < delta_cut (spec Section 3.4)
             t_beta = (2.0 * ALPHA / np.pi) * atanh_factor
             small_t = np.abs(t_beta) < 1e-6
 
             resummed_log = np.empty_like(delta_W)
-            resummed_log[~small_t] = (
-                np.power(delta_W[~small_t], t_beta[~small_t]) - 1.0
-            ) / t_beta[~small_t]
-            resummed_log[small_t] = np.log(delta_W[small_t])
+            resummed_log[:] = np.log(delta_W)
+
+            near_endpoint = delta_W < self.delta_cut
+            if np.any(near_endpoint):
+                resummed_log[near_endpoint] = (
+                    np.power(delta_W[near_endpoint], t_beta[near_endpoint]) - 1.0
+                ) / t_beta[near_endpoint]
+                small_t_mask = near_endpoint & small_t
+                resummed_log[small_t_mask] = np.log(delta_W[small_t_mask])
 
             log_term = np.log(2.0) + resummed_log
         else: