They didn't get rid of MP while keeping TP for any reasons specific to MP or TP themselves. They simply combined Gamewide-Gauge #1 and Gamewide-Gauge #2 into one thing and essentially removed resource concerns from all non-rez-capable classes, with the caveat that, because they're in the same role as Raise-capable classes, certain non-Raise-capable casters will also have to hit a bloat-button on CD (Lucid).
I.e., resource maintenance was purged from the game (replaced with, at most, hit bloat skill on CD) and we were left with some rare cost to using AoE heals in old content (pre-oGCD-heal-bloat) and... effectively... rez charges.
@ Mikey: That said, Mana Shift came at cost to one's own MP and used a % of current MP instantly, while Goad had no cost and gave a flat amount over time. Goad was therefore on paper --and often enough in practice-- far simpler/foolproofed an ability. I can see why one would prefer the design of Mana Shift over Goad.
(@Ath: That just has nothing to do with MP over TP, at least in TP's ARR+ iteration, which was just a duplicate of MP but for non-casters. The last time they were different, even Monks used MP for stances and TP was generated from basic attacks.)
I don't think it'd hugely make a difference if healers and tanks unique outputs were constantly pushed, but... they aren't. And so having alternate direct-contribution spenders available to them for when there's not enough incoming damage to meaningfully leverage their unique outputs seems then to be a clearly good idea.
That said, I don't think it needs to be a black-and-white distinction of role-output vs. damage, nor does it necessarily need to require a additional buttons.
For instance, let's say that --rather realistically-- nuking an enemy as they're winding up a swing might suppress said swing, while nuking them while their defenses are beaten down might cause greater damage. Just with that difference in timing, bankable skills would allow you a degree of control between pure offense and mitigation and therefore your party a lever over risk and consequent reward. With how little suppression can you survive in order to burn the unit down earlier (thereby, say, skipping one of its key skills)? Or, as to be less dependent on bankability (and/or anti-synergy/redundancy between stagger and raidbuffs), allow positionals an impact, with attacks from certain positions at certain times further pushing towards stagger, immediate suppression, or immediate damage/armor-loss/max-HP-reduction (on an self-healing/rezzing enemy).
Or, simply allow for most things to be more naturally hybrids in keeping with their names, themes, and animations -- perhaps even splitting potency output between healing and damage based on how wounded the affected allies are, etc.. Let a CNJ, for instance, have a variety of elemental gauge-spenders such as Quake (more purely suppression and damage, but also able to LoS, knock back, or knock in/collect enemies), Tornado (damaging but also crowd-controlling or shielding against outside projectiles coming in and obscuring allies within), Downpour (slowing enemies but cleansing and --if enhanced-- cumulatively HoTing allies), etc. Let a SCH (or maybe better SGE) overpower a shield to cause it to also reflect damage back at its attacker. Let us leverage what is apparently there so that the effects seem to more naturally flow from theme instead of the theme just remaining a reskin of mostly identical effects.