FEL physics, beyond the basics, gets very complicated very fast; among other things, naive calculations based on the Larmor formula simply don't cut it. In any case, even a shallow reading of the Wikipedia page provides initial answers to your questions, which suggests your question had a far-from-sufficient prior research.
To do some of your homework for you, the FEL radiation wavelength is given by
$$
\lambda_r = \frac{\lambda_u}{2 \gamma^2}(1+K^2) ,
$$
where $\lambda_u$ is the undulator wavelength, $\gamma$ is the electrons' Lorentz contraction factor, and $ K={\frac {\gamma \lambda _{u}}{2\pi \rho }}={\frac {eB_{0}\lambda _{u}}{2\pi m_{e}c}}$ is known as the wiggler strength parameter. The factor of $1/\gamma^2$ is the important one: it tells you that the electrons radiate at the wavelength of the undulator as they perceive it ─ but that this is Lorentz contracted by a huge amount, from the one-meter range down into the subnanometer scales in your question. This is the main tuning knob in the machine: by changing the electrons' energy, the Lorentz contraction changes and so does the output radiation wavelength.
On a separate track, to fill in another apparent gap in your text: the photon energy is entirely independent of the radiated power. The former tells you the size of the energy chunks in the beam, while the later multiplies that by how many photons the beam has to give you the total energy it carries. You seem to be talking about both as if they were the same thing, which makes it impossible to know what you mean by "the value of the energy of the laser flashes".
So, on that line, to clarify:
The photon energy is strictly proportional to the beam's frequency via the Planck relation $E_\mathrm{photon} = \hbar \omega = hc/\lambda$ (so for the upper end of the E-XFEL range, this is on the order of $20\:\mathrm{keV}$, give or take). Thus, this is fixed by the undulator geometry and the electron beam Lorentz factor, as explained above.
The beam power, on the other hand, tells you how much energy is in each pulse (i.e. how many photons per pulse, times the photon energy), divided either by the pulse duration (~100 fs, giving you the mean pulse power) or by the pulse repetition rate (~27 kHz, giving you the average power).
If Larmor physics applied to FEL radiation, then it would apply to the mean pulse power, so it would be completely unrelated to the photon energy.
However, the physics of the Larmor formula do not apply, and it is essentially meaningless in this context. Instead, the beam power is governed by the quantum mechanics of the self-amplified spontaneous emission (SASE) feedback loop, coupled with interference effects between emissions in different parts of the undulator as well as absorption, the electron bunch length and density, the efficiency of microbunching, and so on. (In particular, the energy of each particular electron is pretty irrelevant for this, as the emission is a collective effect.) There simply isn't a clean formula to write it down, and you really need a detailed description of each FEL accelerator and undulator configuration to be able to speak about this.
Moreover, the beam power by itself (which is what Larmor would give you if it applied) really isn't that useful as a figure of merit of the light source; instead, you care about the pulse energy (how many photons per pulse) and the repetition rate, and your ability to focus it down to your needs ─ and this is why the Facts & Figures page quotes the brilliance rather than the beam power.
If you really need raw numbers, they're available at table 5.2.2 of the design report, which puts the (design; i.e. not yet achieved) average power between 65 and 500 W, giving somewhere on the order of $10^{12}$ to $10^{14}$ photons per pulse. However, as I said, unless you're looking to implement an experiment with that beam and you know exactly what you're looking for, the raw numbers on beam power are largely useless.
Further reading: