There is simply no strong enough absorption of any of the greenhouse gases at this wavelength. However this does not mean that there is none.
Remember that what you are seeing here is the transmission function, and plotted on a linear scale, to add to it. The transmission function per wavelength at a certain atmospheric altitude z is defined as $T = I(z)/I_0 = e^{-\tau(z)}$, where the optical depth $\tau$ is the line-of-sight integral over all species' opacities times densities along a path, i.e. $\tau(z)=\int_z^\infty dz\; \sum_{\rm species} \rho_s(z) \kappa_{s}(z)$, where $\rho(z)$ is the density of a species $s$ at a given altitude, and $\kappa_{s}$.
A quick look at the opacity of water at 1 bar and 300K in logscale (taken from the DACE opacity database) shows that there is just a significant drop in the regular molecular wavebands at this wavelength, and nothing else. Same with the other species.
So the fact that just at 4 micron there seems to be a suspicious drop is because our atmospheric column depth integral over $\rho(z)$ at this wavelength just gives a small, but nonzero value for $\tau(z)$. If the transmission plot were logarithmic, you'd see this nonzero value.
Additionally, to try and make sense of
Why is this? It seems like a rather odd coincidence for so many things to be transparent to 4 μm (a specific wavelength of mid-wavelength infrared) EM radiation.
Just think of molecular opacities being periodic signals in wavelength space. If you stack enough periods, you will always find a minimum somewhere, by superposition. For the few ingredients you mentioned, this minimum just happens to be at 4$\mu$m.
