Same as Figure 12, but for the active subsample. We note a deviation from a power law at high EDs in both the observed (first panel) and corrected (second panel) distributions, which was not visible considering the full sample. We attribute this deviation to a sampling bias that becomes noticeable at high EDs when the number of targets is reduced. The uncorrected observational bias at low EDs in the observed distribution leads to a preference for a lognormal fit over a power law, representing the only difference from the conclusions drawn using the full sample. The obtained truncation thresholds obtained are t₁ = 6.0 × 10³² and t₂ = 2.0 × 10³⁵ erg.