Your reasoning to me seems completely biased. Otherwise prove me wrong and give me a list of things WHY he should be stun immune at level 7.
Can you articulate why he should be stun immune at level 10? There's no direct, logical answer to it. It's a trick he's always had, it's a fundamental part of how he plays, it's a unique utility which contrasts Sedna's ability to remove debuffs repeatedly, it's a fun way to save teammates who would be otherwise doomed, especially a fragile one like TB (I've been playing TB/Oak a lot lately), it marks the end of the tedium of the first 7 levels, it's fun. What other answers could there be?
If someone asked me why Oak is overpowered I promise you the first thing that came to mind would NOT be that he gets powerful too fast. Quite frankly that sounds like something someone who doesn't understand Oak at all would say, as he's one of the slowest starting DGs and levels 1-7 are quite tedious.
The reasons why I think Oak is overpowered are that Penitence's self and team damage increase is easily the most effective debuff in the game, Surge of Faith gives his creeps and allies a massive boost umatched by any other DG, and Divine Justice heals for so much that late game Oak is the only DG who can stand in a lane and GAIN health and mana fighting wave after wave of Giants, even catapults offer major mana/health returns. It's a foregone conclusion that his own creep waves will win even with inferior damage and health upgrades, not to mention he's returning that heallth and mana to his entire team.
I just think you went off on a major tangent when you beyond tweaking the bottom three DGs. The idea of nerfing an ability because it protects against interrupts when two of the bottom three DGs don't even have an interrupt (and one is on a significant cooldown and is needed for the basic damage rotation/primary utility) just doesn't make sense to me, it seems more like a potshot at a DG perceived (and rightly so) to be too strong, not a surgical tweak. You're taking a very circuitous and grand approach to fixing what is actually a very narrow problem: Reg, TB, and QoT are too weak.