Actually Helm's Deep would be a perfect map to introduce a wall busting class that like has to put down a bomb and maybe even protect that bomb for a short time before the fuse goes. The book mentions the "fire of Orthanc" multiple times - When it busts a hole in the Deeping wall, when it brings down the gate, and just blasting in the background during the battle.
A sapper class of some sort is definitely a good compromise, but I wouldn't want to see it be quite as powerful as mage or especially torcher could be on v2. That amount of damage blasts dealt combined with limited wood supply would more than likely be enough to ruin the utility that constructor fortifications bring, as you can't just spam patch like you could in v2. (Also, classes with ability to set ground on fire should be handled with caution, given how annoying fire overlay on hud is -- poison classes are annoying, but having to deal with fire hud on top of damage tick is significantly worse imo).
Yes I agree that random classes being able to break the wood faster just because they have an axe is problematic. I need to speak to Cotander and find out how we could solve this issue.
I think that a good way to go about solving the "random axe classes inherently break wood fast" aspect of this problem would be to remove constructor wood from the axe-efficient list of blocks in the server datapack, and explicitly give constructor (and possibly sapper) class axes better breaking efficiency. This would allow these classes to keep the shield disabling functionality while not necessarily being any faster to chop wood.
In 26.1+, there also seems to be a way to set shields to be disabled by certain weapons only, so you could use that modify all the axe weapons to use another item as base, add the tag to all shield classes, and achieve a similar result, but that requires more immediate class modifications, as well as give more work for subsequent class additions given that the constructor/ladderer/sapper niche is almost certainly going to be filled by significantly fewer classes than shield-disabling (it is now by a wide margin, and I don't think the addition of a sapper class to attacker side on every constructor-enabled map would be enough to disrupt that).
A possible solution is to change constructor barricades to a different material like wool. Or add more barricade types made from different materials.
I don't think using wool as a drop-in replacement would work, since the breaking times are much shorter than wood, the block variation is more limited (even in 26.3) so barricades would have to be adjusted, and there'd need to be explanation for why fabric is being used as enemy repelling infrastructure (or trying to convince builders to allow it to be re-textured). Trying anything else as a drop-in replacement would suffer similar problems.
It could possibly function as an alternate option, but I'm hesitant on the idea of making constructor's base state more complex, when I consider it to already be unintuitive compared to how fast a new player is able to select it. Could be like a prestige/specialization/level-up upgrade of the sort that is being considered as one of the point shop use-cases.