Guardedness explanation in manual is inaccurate

[cascremers]

The manual currently explains guardedness only in terms of actions, but omits the case of quantified equations. Notably, these don't need binding to an action, but the quantified variables should only occur on one side of the equation. (Cf. e.g. the definition in Benedikt's thesis)