document that svs.command is an architectural WME

[garfieldnate]

A user's SVS command was not getting run. SVS was active, and the user followed the command structures shown in the manual. The manual doesn't actually show examples of productions creating the commands, however, and it turns out that ^command is an architectural WME, which is not immediately clear from the docs. If the user creates a ^command WME on the SVS WME, SVS will ignore it. I'm not happy with that design, but it would be disrusptive to change.

Update the manual chapter on SVS with an example production that creates an SVS command. Note explicitly that the ^command WME is architectural and should be matched on the LHS, not created in the RHS.

While we're at it, let's also mention in the chapter that you have to enable SVS with svs --enable and can disable it with svs --disable.

[garfieldnate]

While I'm on the SVS section: we need example CLI inputs. See also #490. We also need example inputs to SML's SendSVSInput.