Think of thunderbolt as a recursive technology. That is: Thunderbolt is Thunderbolt + DisplayPort. So TB also is ((TB + DP) + DP) and so on.
The Apple Thunderbolt display separates (de-muxes) the display port signal from the thunderbolt signal (which, as I mentioned above, is a combination of the 2 signals itself). The display port signal is used to drive the TB display, while a clean TB signal (containing it's own display port signal) is passed along to the second monitor.
Because of the way the signal is constructed, it allows for daisy chaining in multiple configurations. You could have Mac - TBDisplay - TBDisplay - External Hub or you could have Mac - External Hub - TBD - TBD or Mac - TBD - ExternalHub - TBD.
However, each thunderbolt device only gets access to one display port signal, which is why it's not possible to have a single dock with dual video outputs. That's why hubs like the CalDigit, even though they have multiple video outputs, can only use 1 at a time. Either the HDMI or DVI port gets the signal.