I voted for this request. I want it too, but it has not happened for a long time. I suspect that this is because there is a catch to implementing it. So without wanting to diminish the relevance of the request, let me explain what the difficulty is, and how I worked around it.
The basic premise of a scene is that it captures the current state of devices, so you can reproduce it. The problem comes when you want to edit an existing scene. Currently it changes the devices to the state that the scene captured. So editing the scene changes things in your home. It activates the scene so you can capture changes to that afterwards.
Now imagine what happens if HA does not change the devices to the existing settings in the scene: Then, when you edit the scene, the scene would capture the current state of the devices and entities in the scene, instead of the state that was previously in the scene. So then the devices in the home change the scene without you touching anything during editing. That would be equally bad, because it immediately ruins the scene upon editing if it captures state again.
So in order to implement this, a scene should just show the settings and allow you to change them by hand, without changing the device and without capturing current state of the device. By doing that, it would move away from the basic premise of capturing state. Then the scene would basically behave the same as a script. But we already have scripts. So why change the scene to what is basically a script, and losing what scenes do best - capturing the current state.
So although I truly dislike editing scenes because they change things in my home, and because editing scenes might capture unexpected changes like motion activated lights, I stopped waiting for developers to think of a way to fix the above dilemma. I started using scripts to define scenes by hand. They can also do more, such as use templates or conditional expressions. By using scripts I do not have to worry that some-one passing a motion sensor while editing a scene will ruin the scene.
I now only use scripts for capturing current states of entities on the fly if I want to restore them to that state later. I never have to edit those scenes anyway, and it is what they are meant to do: capture and restore.
The scenes I have written down in scripts sometimes also use conditionals, or input helpers to set the values I prefer. So they can do more than the scenes.
Maybe the solution to this request is a way to convert a captured scene to a script?