I believe those are currently called "Specialties" and "Backgrounds". The problem here being that there are other things in the D&D-verse that go well beyond what you can/should do with a Specialty. I doubt that many Druid fans would think think that a balanced "Animal Companion" or "Beastmaster" specialty would be anything other than "nerfing" the historically-presented Druid's ability.
Could you design such things as "Super Specialties" which are optional, but balanced between characters taking one?....maybe, but it would be very hard to do that with all the possible modules. Additionally, I don't see how you can add those in while maintaining power level vs published adventure guidelines based on "standard" characters. "If you use Super Specialties, consider your characters as 2 levels higher for the purpose of running published adventures." - always seems to rub people the wrong way.
Somewhat orthogonal to your rumination on published adventures but the general premise is in-line with something I've been thinking on (regarding PC Build Resource Inflation and the "arms-race" implications therein).
With respect to (i) 3e (including the multi-class rules and Prestige Classes) and 4e crowd, the number, breadth and potency of Class Features built into 5e PCs is rather lean. With respect to (ii) Basic, and AD&D1/2 crowd they are slightly more robust. This dichotomy coupled with the interest of making the core game the default experience for much of (but not all...obviously there will be some overlap) the latter crowd creates tension that resonates onward in the design framework. They want the core game playable for that latter market which means that PC output (sans Backgrounds and Specialties) needs to be reasonably in-line with expected monster output and total encounter output (when considering PC group synergy). As such, any power curve growth outside of Class/Race will perturb the PC output vs single monster expectations and the corresponding math that underwrites the encounter formula.
Unfortunately, the former groups' interests are not catered to without extremely robust Class Features and dynamic PC build choices working together to create vast archetype creation capabilities. So you give them Specialties and Backgrounds to further customize their PCs. But is the default combat and non-combat challenge difficulty built with those PC build resources in mind (or are they just going to ask the latter group to handwave it...as they may welcome the opportunity to do so given their appreciation for Swinginess)? What if the current incarnation of Specialties and/or Backgrounds don't carry the payload required to allow players to customize their PCs toward their favored archetypes? The current depth of PC Build Resources within the class structure is insufficient compared to 4e and the Specialties are considerably weaker than 3e Feats. What if they have to buff Specialties and/or Backgrounds (or create a 5th portion of character creation through module - eg; 4e thematic/role powers) in order to make this happen (while still working from the premise that group ii will be playing the default core against default monster and encounter output/difficulty)? In my estimation, they have to eschew the current Specialties construct and buff it to "Super Specialties" level if they want to bring in some of the more powerful (that are well out of line for the current power level of Specialties) Class and Archetype Features that appear to not have residency within the Class Feature framework. What then for group ii? Or what then for monster creation? Everyone outside of group ii buffs default monsters via some sort of template or are they assumed to be fighting a larger sum of default monsters or are the expected to move immediately up a tier in challenge difficulty?
This is where things get tricky and the designers walk a difficult path. Wanting to maintain a non-complex core (and have it relatively balanced versus a default monster/challenge iteration) and then have layers of complexity (and inevitable potency) added on, balanced against a second iteration of monsters or a different encounter formula is a lofty proposition. Further, it will have odd implications for the implied setting in that, effectively, group i and ii may very likely be playing a bit of a different game on a tier by tier basis. That hasn't really happened in the past as there has been decent symmetry within the implied settings of the editions despite their mechanical differences; eg - goblins/kobolds/skeletons/zombies > orcs/gnolls/hobgoblins/specters/wights > ogres/ghouls/wraiths > giants/vampires > mind-flayers/beholders/death knights > liches/dragons. Does group i forgo fighting goblins, kobolds, skeletons, and wraiths? Do they just fight many of them (as Bounded Accuracy allows for this)? Do they add a template or passive buff to all creatures?
The other idea they're throwing out I find interesting but, again, prone to gross issues of power imbalance without extraordinary QC. Treating a feature or build resource (such as a companion) as the power level of another PC in the group (toward the encounter balance equation) demands power in accords with;
- the intra-combat suite of actions available to a standard player of a standard class.
- in-line breadth and potency of that suite of actions.
- aggregate survivability modification of the group due to the total HP inflation/damage soak capabilities that another PC would afford.
- the positional/tactical augmentation of having another friendly creature occupying a space of the arena of combat.
- the extra-combat resources that another class brings to bear.
Boy, that is an awful lot of power for a single class feature and an awful lot of room for intra-feature power imbalance (especially if they decide against augmenting specialties and using this formula for other powerful features...features that don't count as another space-occupying creature).
Due to the feedback loop of adjusting one aspect of PC build resources and how that affects PC output versus expected monster output versus total encounter output, the quality control effort will need to be extraordinary (for what effectively seems to be a small business - a small franchise of a large corporation that must make a case for its earning potential and legitimize its expense ratios to its shareholders). Bounded accuracy will help them a bit (as they will not have to deal with the 2nd and 3rd order functions of to-hit percentages) but there are so many variables beyond "to-hit" that I'm hesitant to hope for each system (core and the various customized iterations)...especially without group ii having to fiddle with the default assumptions of monsters and/or the encounter formula.