The reason to build a Dyson Sphere (or Swarm) is that you want all (or at least a large fraction) of the energy output of a star. To "build decentralized fusion reactors" that can provide the same scale of energy is even less practical than building a shell around a star and would require far more materials! Also, fusion is really simple when the ignition energy is provided for free by the gravitational compression of something the size of a star, and not so simple when you're trying to get it started on a small scale using any other form of energy for ignition. The bottom line is we don't really know if small-scale, controlled, net-energy-positive fusion is possible at all, but if it is it has a lot of overhead costs... you then have to deal with ignition energy, containment, etc. You're trying to make a mini-star and keep it tame. The physics are not favorable to this, they are favorable to star-sized stars, where gravity and fusion energy pressure can balance each other for millions of years.
The concentrated energy of Dyson Sphere can be very useful. Feed some of it into lasers and then launch starships and then slow them down when they get to destination. Not tiny probes with fusion reactors but full size starships.
Feed most of the energy of star to lasers and end up with weapon that will melt planets across the galaxy.
Honestly, don't need to expand across the galaxy if have Dyson Sphere, which could be explanation of Fermi Paradox.
Diffraction limit determines the fundamental beam divergence angle theta = lamda/pi/D, where D is the beam diameter, lamda is wavelegth of elwctromagnetic radiation (eg. light).
To minimize theta, we need to either increase D or decrease lamda.
Lets assume we would be able to make far infrared high power lasers, at say 10,000 nm = 1e-5 m wavelength.
Lets assume we would be able to make D, the diameter of our laser beam, similar to the diameter of a typical planet, for Earth it is ~13,000 km = 13e6 m.
Theta = 1e-5 / 3.14 / 13e6 ~= 1e-13 radians.
Sun is ~ 25,000 light years from the center of our galaxy, ~= 25e3 y 3e8 m/s 31.5e6 s/year ~= 1e20 m.
Laser beam diameter, there far away, would be: 1e-13 *1e20 = 1e7 m, similar to the diameter of Earth, not much further diverged, focused and delivering the wast amount of energy all over the planet thereby evaporating it to a gas.
Indeed, what you say about the melting far away planets is possible, in theory.
The scale.
The reason to build a Dyson Sphere (or Swarm) is that you want all (or at least a large fraction) of the energy output of a star. To "build decentralized fusion reactors" that can provide the same scale of energy is even less practical than building a shell around a star and would require far more materials! Also, fusion is really simple when the ignition energy is provided for free by the gravitational compression of something the size of a star, and not so simple when you're trying to get it started on a small scale using any other form of energy for ignition. The bottom line is we don't really know if small-scale, controlled, net-energy-positive fusion is possible at all, but if it is it has a lot of overhead costs... you then have to deal with ignition energy, containment, etc. You're trying to make a mini-star and keep it tame. The physics are not favorable to this, they are favorable to star-sized stars, where gravity and fusion energy pressure can balance each other for millions of years.