I'm not sure I like the idea of adding a whole extra layer of complexity and type memorization, but I like the creativity, and your example of a situationally useful Racoon.
I think Racoon's whole point is to be a beast in a dedicated Ninja and/or Cute Animal deck. The problem is that he doesn't do that: unattackable just means that he's worthless on defense; unstoppable with 3 atk isn't great unless you have a way to boost damage... which you don't, because you spent those tech slots loading up with ninjas so that you'd have a chance of getting the ability to happen in the first place (and to protect the Racoon on the back line).
Basically, for that specific example, I think making Racoon a 2-cost would be much more effective at getting it to do what it wants to do. Making it 5 ATK (keeping the 4-cost) would be even better, because then it becomes: if you have a ninja, this eats a tech building or a hero each turn. And that's what a real racoon would do: get into your house and knock over trash cans.
If you want something specifically good against Hyperion, you go for Glorious Ninja. It's just that GN is also specifically good against almost everything else. But with a 5-ATK MR, you might choose to have one fewer GNs to put a MR on attack.
And I'm not sure any amount of type bonus would make Arresting Constable or Debilitator Alpha worth it at their costs.
But I really like the point you're making about underused cards. Fore every underused card, there should be a situation where you can go, "I think my opponent will use that known and solid strategy (out of his several options); this card will make that strategy weak!"
So the question becomes, "What specific strategy is this card designed to beat?"
* Racoon seems designed to beat a wall of units or maybe to survive a red-style barrage and counter-attack? The problem is that Raccoon is so expensive that you won't have enough units in play to get him to do his thing, even against Tech 0 and Tech I armies.
* Arresting Constable seems designed to take out much bigger cost-heavy units or units with killer abilities. The problems are that he's as or nearly as expensive as many of those units, he's too weak to survive to use his ability, and he can't arrest Tech III units (the ones with the biggest stats and killer-est abilities). He'd be great against fliers/OP units, if so many of them didn't have haste. And he takes a slot away from actually useful units. (I think he should either be 1-cost or 5/5.)
* Debilitator Alpha seems designed to counter the very specific (and serious) threat of Tiny Basilisk. Which would be fine if DA were a Tech I unit that was a little weaker and a lot cheaper.
Sorry to rant!
It's an interesting idea; maybe others find it more appealing than I do. (: