Is a symmetric matter-antimatter universe excluded?

We consider a non-standard cosmological model in which the universe contains as much matter as antimatter on large scales and presents a local baryon asymmetry. A key ingredient in our approach is that the baryon density distribution follows Gaussian fluctuations around a null value $\eta = 0$. Spatial domains featuring a positive (resp. negative) baryonic density value constitute regions dominated by matter (resp. antimatter). At the domains' annihilation interface, the typical density is going smoothly to zero, rather than following an abrupt step as assumed in previous symetric matter-antimatter models. As a consequence, the Cosmic Diffuse Gamma Background produced by annihilation is drastically reduced, allowing to easily pass COMPTEL's measurements limits. Similarly the Compton $y$ distorsion and CMB 'ribbons' are lowered by an appreciable factor. Therefore this model essentially escape previous constrainst on symetric matter-antimatter models. However, we produce an estimation of the CMB temperature fluctuations that would result from this model and confront it to data acquired from the Planck satellite. We construct a angular power spectrum in $\delta T / T_{CMB}$ assuming is can be approximated as an average of $C_\ell$ over a Gaussian distribution of $\Omega_B$ using Lewis & Challinor's CAMB software. The resulting $C_\ell$ are qualitatively satisfying. We quantify the goodness of fit using a simple $\chi^2$ test. We consider two distinct scenarios in which the fluctuations on $\Omega_B$ are compensated by fluctuations on $\Omega_{CDM}$ to assure a spatially flat $\Omega_\kappa = 0$ universe or not. In both cases, out best fit have $\Delta \chi^2 \gtrsim 2400$ (with respect to a fiducial $\Lambda$CDM model), empirically excluding our model by several tens of standard deviations.