I'm more on the side of adding only a minimal number of items—just the ones that really are the missing counterpart to existing elements, or in any other way fill an important niche.
Take a look in the editor at the tabs with all game elements. Getting pretty crowded in there, not? With TCB we have huge diversity in game elements and together they create potential for the most complex puzzles one can think of. However since the days of AE with the introduction of more and more of these elements, people have slowly forgotten what was possible with the original minimal set of features.
That is not to say that I don't like the additions from JtRH and TCB. Most are great indeed, like the additional snakes and tarstuffs, the scripting engine… can't name them all. However, not so much with JtRH but certainly with TCB, there have been added some less successful elements as well (I'm thinking stalwarts, builders). They are more interesting gimmicks, instead of powerful elements that integrate well with the game.
Also, some elements have been added, that although I think are great, were not really necessary to make an excellent game—I would have loved DROD equally without them. Take for example the addition of the Fegundo in TCB. Although it's a pretty neat monster and I can see its potential, it doesn't really fill a niche in my opinion. Its mechanics are not really comparable with anything else in the game. It takes a far greater step away form the existing mechanics compared to a new tarstuff like gel, which works mostly the same as other tarstuffs, but has new cutting rules.
In other words: to fill niches you'll have to look at which properties of elements are comparable and in what combinations they are missing, not just add an element with entirely different rules and say it's the missing link 'because there wasn't anything like it yet'.
So then what things do I like to see in TSS?
A few things which I think would fit in well are diagonal squares to complement the orthosquares, serpent switch tokens similar to the tarstuff switchers…
The monster that I always missed was one that moves diagonal only, but since I scripted the salamander it's not on top of my wish list anymore (still official support would be neat). Regarding scripting, with the popularity of creating custom monsters it would be nice to be able to retrieve coordinates of the player and NPC's; make commands like appear at more flexible by letting them take variables as arguments.
I'm also going to second what Blondbeard said: keep it simple! All the good elements have fairly simple properties, and are because of similarities between them intuitive and easy to remember. Now if the Caravel team sticks to the niche filling philosophy any new elements should automatically follow this advise.
Nevertheless it's a useful thing to keep in mind. Better not to try and create overly complicated bosses this time; for example the slayers have lost quite a bit of their appeal now that people have found the many ways to kill them.
All in all, we don't need many more elements to make more complex, versatile and generally more enjoyable holds, the existing features are very well capable of that. Al we need to do is round up the set of elements neatly for the final installment.
____________________________
Holds:
An Architects Audition,
Artful Architecture,
Salamander,
Elusive Exhibitions,
Leftover Levels,
Six Times Six
Collaborative:
Way Forward,
Advanced Concepts 2
Styles/Mods:
Basalt,
Sandstone,
Garden,
Clock using game elements
is a considerably more stylish way to express sarcasm than 
, because everybody uses 
