Theoretical and observational approaches to settle the important questions of the progenitor systems and the explosion mechanism of normal Type Ia supernovae thus far have failed. With its unique capability to obtain continuous spectra through the near- and mid-Infrared, JWST now offers completely new insights into Type Ia supernovae. In particular, observing them in the nebular phase allows us to directly see the central ejecta and thereby constrain the explosion mechanism. We aim to understand and quantify differences in the structure and composition of the central ejecta of various Type Ia supernova explosion models. We cover the currently most popular explosion scenarios using self-consistent multi-dimensional explosion simulations of delayed-detonation and pulsationally-assisted gravitationally-confined delayed detonation Chankdrasekhar-mass models, and double detonation sub-Chandrasekhar-mass and violent merger models. We focus on the distribution of radioactive and stable nickel in the final ejecta in density and velocity space. Of all models we study, we find that only in the double detonation scenario with a surviving companion the inner ejecta are anywhere close to spherical symmetry. It is thus crucial to simulate in particular Chandrasekhar-mass explosion models in 3D, because the lack of buoyancy in 1D models leads to an unphysical inverted stratification of radioactive and stable nickel in 1D models. The ejecta of Chandrasekhar-mass and sub-Chandrasekhar-mass merger models with an exploding secondary white dwarf are inherently 3D. In all of them nickel is distributed over a wide range of densities even at fixed velocity. Therefore, it is crucial to compute synthetic observables in the nebular phase, when the ejecta are optically thin, in 3D from the full 3D ejecta, because spherical averaging instead leads to unphysical ejecta properties. (abridged)