\section{Normalization of the published bounds} \label{app:norm} The conversion~\eqref{eq:convert} needs the normalization $C_{\rm pub}$ of each bound in Table~\ref{tab:bounds}, measured against Di\'{o}si's master equation, \begin{equation} \dot{\hat\rho}=-\frac{C\,G}{2\hbar}\iint d^3x\,d^3y\, \frac{[\hat M(\mathbf{x}),[\hat M(\mathbf{y}),\hat\rho]]}{|\mathbf{x}-\mathbf{y}|}, \qquad C=1~\text{(Di\'{o}si)} , \label{eq:diosi-me} \end{equation} with $\hat M$ the Gaussian-smeared mass density. Equivalently, the noise is that of Eq.~\eqref{eq:dp-noise} and the energy functional of Eq.~\eqref{eq:dp-rate} carries the prefactor $G/2$. \paragraph{Penrose normalization, $C_{\rm pub}=8\pi$.} Donadi \emph{et al.}~\cite{Donadi2021} write the master equation with prefactor $4\pi G/\hbar$ in place of $G/(2\hbar)$, a ratio $8\pi$, and their Supplementary Information states that their rates agree with Di\'{o}si's apart from that factor. The MAJORANA analysis~\cite{Arnquist2022,Arnquist2023Erratum} uses the same convention, which we check from its own emission formula. At $C=1$ the acceleration $\mathbf{a}=-\nabla\Phi$ of a point charge in the smeared noise is white, with $\langle a_ia_j\rangle=D\,\delta_{ij}\delta(t-s)/3$ and \begin{equation} D=\int\!\frac{d^3k}{(2\pi)^3}\,k^2\,\frac{4\pi\hbar G}{k^2}\,e^{-k^2R_0^2} =\frac{\hbar G}{2\sqrt\pi\,R_0^3}. \label{eq:accel-noise} \end{equation} Larmor emission from this acceleration gives, per charge $e$, \begin{equation} \frac{d\Gamma}{dE}=\frac{e^2D}{6\pi^2\varepsilon_0c^3\hbar E} =\frac{1}{12\pi^{5/2}}\,\frac{e^2G}{\varepsilon_0c^3R_0^3E}. \label{eq:larmor} \end{equation} The corresponding coefficient in MAJORANA's rate formula, per $Z^2$, is $2/(3\pi^{3/2})$, which is $8\pi$ times $1/(12\pi^{5/2})$; the factor-of-five statement checked in Section~\ref{sec:survivor} confirms the reading. \paragraph{Di\'{o}si normalization, $C_{\rm pub}=1$.} XENONnT~\cite{XENONnT2026} states that it follows the Di\'{o}si as opposed to the Penrose convention, which differ by $8\pi$. Vinante and Ulbricht~\cite{VinanteUlbricht2021} give the heating per particle as $G\hbar m/(4\sqrt\pi R_0^3)$, which is Eq.~\eqref{eq:heating} at $C=1$. For a uniform sphere at large separation, Howl, Penrose and Fuentes write the Penrose time as $5\hbar R/(6GM^2)$~\cite{HowlPenroseFuentes2019}, which is $\hbar/\EG$ with Eq.~\eqref{eq:gamma-sat} at $C=1$. For the LISA Pathfinder row we take $C_{\rm pub}=1$; recomputing the bound of Ref.~\cite{Helou2017} from its force-noise formula and data with that reading reproduces its $40.1$~fm. The force-noise rows sit three orders of magnitude below XENONnT, so a factor $8\pi$ would not change the binding bound. \paragraph{Tilloy and Stace, $C_{\rm pub}=1/2$.} Ref.~\cite{TilloyStace2019} uses the heating coefficient $G\hbar m/(8\sqrt\pi r_c^3)$, half of Eq.~\eqref{eq:heating}. Its bound is an estimate for the conservative surface temperature of PSR J1840$-$1419 with unit emissivity, not a confidence limit.