Thoughts about the Manabound-

Overall-
Fluff: Very neat concept. Sorcerer, but on purpose.
Mechanical: Perhaps a bit too much overlap with the paladin which I expect would lead to unfavorable comparisons. I really like the hit dice interactions.

Specifics-
Name: I don't love it. I don't think that D&D ever uses the term "mana" in this way and IMO it doesn't fit well. Something like "Channeler" would work better. Yes, that's a very generic name, but that's kind of the point. This is a matter of taste, however.

Hit dice: I'm going to say this should be 1d10. The class seems built to be in melee and doesn't have much to enable skirmishing. And it's too MAD to have a decent investment in CON
Light Armor Proficiency: I'm not sure why they have this. Mana Surge makes a dex build pretty suboptimal and the Unarmored Defense feature makes it kind of moot anyway.
Tool proficiency: I don't have an objection to this per se, but I don't really understand what the justification for it is in terms of the lore of the class.

Features-
Spellcasting: No major objections, except that the lack of access to spell foci will become annoying. Choosing cha as the casting stat is a very safe choice, but I'm going to step out on a limb and suggest that this might be the perfect class to be a con-based caster. It would settle several of my other objections and it would be different kind of thing than we've seen before.

Unarmored Defense: Compares favorably to other versions of the feature, except that the Manabound is already struggling to juggle its stats unlike the barbarian, and unlike the monk neither of the armor stats are an attack stat. Doubly problematic due to the 1d8 hit dice and lack of features that let you weave in and out of combat. I expect most manaborn would see a dip into a class with heavy armor proficiency as mandatory.

Consume Self: I love it. It's multiclass bait, and so maybe should be pushed back some, but it's a really cool feature.

Mana Surge: This is a very cool feature, but I feel like it's extremely restrictive in terms of what spells they can use for this. For example, a gravebound has only cause fear as a viable option for this right up to 9th level. The arcane magus can't use this feature at all until 5th level. I know you're using this restriction to try to limit Ravnica/Eberron extra spells shenanigans, but I think you're better off giving the players more options, even if some options turn out stronger than intended, rather than locking up gameplay to a single viable spell for certain characters.

One with Magic: I have an irrational dislike of d4s, so I'm likely not very objective on this, but I see no reason this shouldn't be a d6.

Seize the Threads: a lesser-wish kind of a deal? Seems okay. Did you mean "no material components" or "no costly material components"? Because if the spell can't have any material components at all that's a pretty major limitation on spell selection. I would actually probably rather see something more like "Magical Secrets" where they get a list-free spell pick, and maybe can replace it during a long rest or something, but I feel like a totally spontaneous spell pick would lead to a lot of book flipping mid-session for some players.

Subclasses-
Gravebound: This looks good, but this is in no small part because it solves some of the issues I have with the base class. Spell selection is rough when it comes to spells viable for Mana Surge until 9th level, as noted above.

Faithbound: Spells have plenty of solid options for Mana Surge, so that's good. You solve the "Cure wounds is more valuable than a hit die" problem by giving them more things to do with hit dice, which is elegant. My favorite subclass.

Arboreal Guardian: Issues where at 3rd leave your only options for Mana Surge spells are spells that end when an enemy passes a save which compares poorly to the Faithbound's access to Bless. 5th level solves the problem though.

Arcane Magus: No Mana Surge spells available at all at 3rd level. Arcane disjunction is very cool. If you fix the Mana Surge issue this is my second favorite subclass.