The Sequencer is one of the meta-modules that create signal flow structures and, in this case, generates KeyEvents which are vectors, rather than individual continuous controls. The Sequencer creates a polyphonic (potentially overlapping) stream of vectors of (!KeyDown !KeyPitch, !KeyVelocity, !KeyTimbre). It's designed so you can take any Sound that was set up to be controlled by a keyboard and instead control it with the Sequencer (or a MIDIVoice with File or Script as the source). When polyphonic, Sequencer (and MIDIVoice) also create multiple copies of the signal flow to their left and handle the voice-stealing policy.
As an environment, the Sequencer also "binds" the KeyEvents to specific values so they are no longer seen as variables to any Sounds to the right of the Sequencer in the signal flow.
To the keyboard-controlled Sound, it looks the same as if those KeyEvents were coming from an external keyboard (even though they are being algorithmically generated).
In order for the Sequencer to be a control source (rather than a meta-module that creates an environment as it is now), it would have to have multiple outputs: an output for !KeyDown, an output for !KeyPitch, and output for !KeyVelocity and an output for !KeyTimbre. The control source would not be able to create multiple copies of the Sound it was controlling so you would not be able to specify polyphony. The Sequencer would have to be pasted into each parameter of the (formerly keyboard-controlled) Sound, making it tedious to switch between sequencer control and external keyboard control (or MIDIVoice Script or File control).