New Spell to Increase High Level Attack Variety

Problem

What clash of clans does well at, and can make better, is attack variety. Top strategies include: barch, giant healer, mass dragon, loonian/balloonian, and gowipe. However, the giant healer strategy is not as effective at higher levels. Perhaps the L7 giant update will fix this, but if it does not, I have a new proposal that may not just fix this strategy, but introduce new strategies.

The Idea

That is, a new spell: scatter spell. Or whatever. That's just what it does, drop it on some troops, and they scatter towards different targets. The exact implementation of this is in the details, but the idea is that near the end of a raid, a weakness of things like giants is that they are cluttered and slow. If they were split off into groups, they could stand a better chance of finishing the job.

The Algorithm

This may, or may not, seem simple, but it is in fact pretty complex on the details to implement. If you are just thinking about that one use case, your algorithm might have weird consequences if using it in the middle of a base. Thinking through this, I think random scattering, close building targeting scattering, clustered scattering are flawed. The way to do this to both expand the use case of the spell while simultaneously making it more user friendly to predict behavior is to make it a toggle spell. Essentially, you drop the spell on some troops and they will continue to scatter away from their average center until it is toggled off. Thus, there would be no duration, the radius would only be important at drop time, and the troops would continue scattering until otherwise commanded. In this way, someone who is using the spell near the end of a raid will let the troops scatter for a while, while someone else might want it to scatter troops after breaching a choke point.

The only problem I see now is that troops could potentially scatter in directions where there are no buildings, and this is obviously undesired. Adding directed scatter behaviour on top should fix the problem.

Discuss

Implementation details - cost, time and town hall level that is requisite
Is this two powerful? In what way could it be modified if that where the case?
Other uses - perhaps scattering goblins for a short time after breaking a wall, or balloons/dragons