Question 1: There are really surprisingly few "standard libraries" or tools for this kind of thing. You would think we'd have a CA editor capable of doing object-based editing by this time -- like, copy in a complete device made out of reflectors and converters, each of which is made out of still lifes and oscillators, each of which is made out of cells, and you could do "group" and "ungroup" operations and snap to the right locations to fit the circuits together correctly.

But at the moment, pretty much all we have is tools to copy and paste rectangular sections of patterns at the cell level -- plus we've got good scripting tools (in Golly) that can be used to string together whatever pieces we might want, but it's up to individual pattern-builders to write those scripts for each specific purpose.

So our "library" is pretty much just the LifeWiki and a few other pattern repositories, and we borrow liberally from existing large constructions -- but when we're building something new, we usually just build flat bitmaps, not anything with built-in annotations or metadata.

Question 2: The thing that's been the most interesting to me in the last decade or so is the increase in collaboration. Projects used to be done by just one person more often than not -- but now a very large fraction of the biggest discoveries are completed via a large group effort over the course of a few weeks or month. One big recent example has been the RCT fixed-cost universal glider synthesis project, which needed contributions from quite a few people to solve all of the tricky little sub-problems:

  https://conwaylife.com/wiki/Reverse_caber-tosser