Every big AAA game that uses Unreal/Cry/Unity will still have their own internal engine programming team, staffed with tech programmers who's job is to modify the engine to suit the needs of the game.
They'll also have at least a dozen talented gameplay programmers, plus specialists like effects programmers, network programmers, audio programmers, etc...
All that buying one of these engines does for them is save them *time*. Instead of their engine team spending a year making their own engine from scratch before the game's production begins, they can start almost right away.

Let's say i start my own first-person-shooter company.
Either:
A) I hire half a dozen programmers and a tech-artist and we spend a year writing our engine. Then I hire half a dozen more programmers plus a dozen artists and we spend another year making our game.
B) I hire half a dozen programmers and a tech-artis and we spend a few months modifying an existing engine. Then I hire half a dozen more programmers plus a dozen artists and we spend a year making our game.

Same staff required. Same outcome. Only difference is time (which in business, means money).

since the invention of the PC, games have evolved from text mode graphics and pc speaker sounds to multimedia 3D graphics apps with 3D/surround sound, voiceovers, music, etc.

as such, the talents required have grown from just coding, to coding, artwork, sound work, writing, motion capture acting, etc.

so while coding was the only skill originally required to make a state of the art text mode game in the early 80's, nowdays its just one of the basic skills: design (aesthetics, dynamics, mechanics, level design), coding (engine,special effects, AI) artwork (meshes, textures, animations), sound (voice, music, sfx), writing( quests, dialog, backstory), etc.

so coding is no longer the only skill required, but its still required.

once you have an engine or library (3rd party or inhouse), it looks like coding becomes a smaller percentage of the overall work compared to things like creating graphics assets. at least that's what i'm discovering.

coding will never go away because tools to make games without code can only do what they're designed to do. if you want anything else, you have to code. so if you want to go above and beyond what everyone else is doing with the same tool, you have to code.

FYI: as a coder first, i too somewhat lament the fact that other skills are also required - unless you're cloning dwarf fortress or something. But at the end of the day, i'm a gamedev. coding is just a means to an end. so nowadays i spend most of my time using blender.

It's also worth noting that as an indie developer, your idea is the critical landmark! Have a good enough idea and

everything else will follow :D

Just because a game can be created with all these glorious graphical enhancements, doesn't mean it's good. All

the Crysis games have amazing graphics and scope, but pitch that against Terraria, Tetris, Mario, Megaman, or whatever,

and I'd pick the classics any day :D

I'd also imagine that no matter how incredibly well designed an engine is, there will ALWAYS (99.99%) of the time, be something

that was overlooked that will need manually putting in. It's practically impossible to design something perfectly from the offset.

since the invention of the PC, games have evolved from text mode graphics and pc speaker sounds

And the only thing that prevented people from playing games on punch-card machines was the tangible, measurable, physical cost of performing I/O.

in 1978 as a sophomore in high school, i wrote a turn based version of the arcade machine game "lunar lander" in basic for the ibm 360 mainframe. we had 3 dumb terminals and a line printer terminal running at 1200 baud over 4 dedicated copper phone lines to the county's mainframe. like many simple text mode turn based games, it used the trick of drawing a whole screen of text at once to refresh the scene. so you'd enter your move, then it would draw a whole screen and prompt for your next move. i played it on the line printer one day and generated a printout over 50 feet long, one page per screen.