What makes Pelennor different is the fact that, from the point that Aragorn recruits the ghosts prior to the battle, the protagonists have a bigger army and the conflict is non-existant. The problem posed is: 'How do we win against a bigger army?' and the solution is 'We actually had a bigger army all along.' The ghosts are free to show up at any point, because there's no constraint attached, and so there is no real conflict. What makes Helm's Deep different is that the critical condition set up before the fight is 'If you hold out for five days, you will win.' The tension comes from whether or not they can hold out for long enough.
The other uses of Azem's crystal have been as a ludonarrative device to explain the duty finder/party finder. The primary conflict at the climax of 5.0 is really centered around you and Ardbert. By the time that G'raha makes a move you've already won from a plot perspective; the only thing that remains is to queue up and find seven other party members to fight Emet. You can take G'raha out from the picture completely and the conflict's resolution remains unchanged. A better example for you would have been to reference the Seat of Sacrifice, where you counter Deus Ex Machina with Deus Ex Machina. Elidibus arbitrarily banishes you to another dimension on a whim (why not just do this earlier), and then you arbitrarily use Azem's soul crystal to summon Emet and summon you back. You can say that that sequence was a bit underwhelming, but it's really just a ludonarrative device to set up a quicktime event. I think that if the soul crystal gets used for routine plot problem solving, then they'll have to come up with some more restrictions on its use.
Plot armor is a different issue entirely. The reason why writers are reluctant to sacrifice major characters is because they have story potential. You can do it at any point for the sake of easy drama and the name of 'realism', but you're trading a short term gain for a long term loss. If you're going to do it, you have to be smart about it. They still can kill off any Scion that they like. It's just a question of what kind of a story they want to tell through it. I think that had Zenos killed off Y'shtola, you would have automatically forced him into being this season's primary antagonist. I think that route was open to them when they were initially writing Stormblood, but I think that the end result that we have here was a better writing decision overall.
From a season 2 standpoint, if they wanted to go with a front and center primary villain (perhaps the vengeful soul of the Warrior of Light who was killed on the 13th by Igeyorhm?) then you'd have a good reason to kill off a Scion in order to set up a grudge (a common enough writing decision in a Final Fantasy game). But there's no point in killing major characters off for the sake of killing them off. The writer needs a good reason for doing so.