The way I view scenes and modes is that modes are there to indicate a state that the house is in and this state can be used across any number of "smarts" within your home, being used to drive behaviour in rules, in apps, etc. Scenes on the other hand, I see as pre-configured settings for, typically, a collection of devices, where you want the ability to quickly and easily adjust them all at once and be able to refer to a scene in things such as a rule, rather than configuring the settings for a scene in multiple places, making any changes to the scene more time consuming.
When using them together I would have modes drive the use of scenes, not the other way around, e.g. a transition from one mode to another would trigger the activation of a scene, where that scene may include lighting, A/V equipment, thermostat settings, etc. I don't use scenes configured in HE myself, but if what you are saying is correct, then think of my comments relating to scenes in a general sense.
In terms of triggering, my suggestion would be 3 separate levels.
First would be setting up a virtual switch used for external triggering, such as a Google Routine. I would include some reference in the name of the switch to say it changes the mode to Night. Also, set the switch to turn off automatically after a few seconds.
Second, I would setup a rule that is triggered by the virtual switch, when triggered this rule changes the mode to Night, in this case.
Third, I would setup a second rule that is triggered by the mode changing to Night. When triggered the rule will activate the scene you want, define in HE scenes, and anything else not covered in HE scenes you add as extra actions, such as your thermostat change, turning off some smart plugs, A/V equipment, etc.
Obviously you could also evolve this approach to have the rules handle changes to different modes, the virtual switch could be a toggle between night and day, etc.
I have this setup (minus the HE scene) for a few reasons, but the main one is that it allows for more than one trigger point to initiate the mode change. You could place a mode tile on a dashboard as a method of changing the mode, bypassing the first rule, but still having the same effect by triggering the second rule. Equally you can also have a physical button, another rule, etc that could also trigger the mode change.
I agree with you, if HE scenes don't include other devices it would be a nice feature to expand it, just not the mode part, IMO.