I'm always surprised on how uneven and hacked together the surface of these shuttles look. I'm pretty sure there's a solid reason for keeping it that way, one would just think it's easier to spot issues on the surface if it's more consistent like the surface of a plane.
We are all used to the smooth and refined surfaces, assembly, PCBs, etc. of consumer products. Things like the Discovery are unique, there is no real manufacturing process built around them. A passenger plane looks much less 'hacked together', but it's still a bit hacked compared to a Macbook Pro. This is because of the difference in the numbers: 1 vs 1000 vs 1000000 pieces.
Also, this is constantly repaired, updated, etc., not just thrown away and replaced by another one.
The number on the surface of a new tile is more or less the filename of a CNC gcode program that can create a new tile to a precision quite a bit higher than you can see visually... You can read a tangentially related discussion of this here.
There's at least 20K HSRI "model numbers" and over the years probably many more as revisions were made, they made lots of Q+A test pieces and there were many revisions and lots of repairs and there's a lot more than just black HRSI tiles, I'd estimate somewhere in the realm of "way more than 10M tiles" were made over the years. The hotter they get the smaller and thicker they have to be, something about differential thermal expansion stresses or something.
The TPS as a whole was quite a technological accomplishment.
Consumer products are designed to be sold where Discovery and commercial aircraft where designed to be used. Occasionally consumer products will evoke that style such pre-linovo IBM laptops. Corporate jet's are the other side of the coin where company's focus on style not just function so dispite low production runs there far sleeker than 140 Boeing 787 Dreamliner out there.
The tiles I've seen (including physically touching) have been smooth as a countertop, at least when new.
I believe this pix was taken in the post-FIB era. Flexible insulation blanket. Not all tiles could be replaced by FIB but some could and were. I imagine the heat load was not very high in those positions, so FIB makes sense. And I know FIB looks exactly like the wavy blanket quilt look you're talking about and is seen in the pix. FIB had another name which I forget. The nomex based "frizzie" (FRSI?) looked less quilt-y, more like a prom dress fabric, well, kinda. I don't see anything I recognize as frizzie in the linked pix. Frizzie looks like the inside layer of a firemans coat, which also probably doesn't help describe it very much, yet, there it is.
TLDR is that wavy tile is not a tile, its a quilt.
Most people know the engines were under continuous development for practically the whole program; well, so was the heat shield system. I can't be bothered to look it up and it doesn't matter anyway, but I don't think the same TPS configuration ever flew more than three or so times in a row without some new idea being tried, at least in the early decades. Continuous integration.
Personally I always suspected we'd lose more orbiters due to continuous integration fiascos than mechanical damage. Then again they spent a lot of money on research, Q+A, and experimentation so what would be a disaster in a typical IT shop works pretty well in aerospace.
On reentry the tiles burn and degrade, and thus had to be tested and replaced on Earth if necessary. They are all numbered with serial numbers that uniquely identifies them and their location on the surface of the shuttle [1]. Like a real life hash table ;)
Smaller tiles are easier and cheaper to replace, and thus the uneven look.
I think the unevennes is due to the replacability of the tiles, which is how the tiles end up being of different manufacturing time and degree of wear.
Could very well be functional. The rippling looks to me similar to the rippling and unevenness of the SR-71, which wasn't smooth because a smooth surface would be less stable at speed.
Actually, portions of the SR-71 skin were corrugated, much like the older (and much slower) Ford Tri-Motor, because the high temperatures of Mach 3+ flight caused the skin to expand and contract dramatically throughout the flight profile. This corrugation was actually against the wishes of the aerodynamicists, who would have preferred a smooth surface.
