Since a lot of the answers are perspective-driven rather than evidence-driven, I will give my own perspective and then accept the most popular answer.
First off though, I wanted to address this comment I made in the question:
I'm looking for answers that specifically target teaching rather than research presentations. (For me, there is a significant difference.)
The difference is that for research talks, the meat of the content is in the paper already. The talk is an advertisement: a flavour of the content in the paper. When I prepare research talks, I'm quite happy to keep things at a high level of detail and to avoid bullets almost entirely and let those few people in the audience who are interested in more detail defer to the paper.
For teaching, the meat of the content may be elsewhere: it may be in a book or it may be in handouts. The level of textual detail in the slides does depend on a balance of what other material is made available alongside the slides, what can be noted in the talk, etc. For example, I'm currently preparing a course on a new area that is without an authoritative text-book; I'm preparing the course as I teach it. Though I could provide separate handouts, I prefer to use the time I have to capture the content in the talk/slides. Currently my strategy is to write text-heavy notes and annotations into each slide (where needed) to give the context and some description for the slide. This is lower-effort for me, meaning I can keep the slides for introducing topics and keep everything together, and the students can refer to the slides as handouts without being hit over the head with text during the talk.
Also, the idea of using (only) the board is in principle good for courses that are, e.g., mathematical in nature, and where teaching process is important. However, using the board is far too slow for subjects that wish to cover a lot of factual knowledge. For example, if I need to teach about the table of elements, I'm not going to draw that on the board. (Furthermore, I remember in many a board-heavy lecture as an undergrad, concentrating solely on transcribing everything to read later ... I think a lecture where a student is constantly writing is a bad lecture.)
Anyways, back to answering my own questions (parts of my answers are indeed informed by/similar to other opinions):
Should bullets generally be introduced one-by-one using animations or should the entire slide be displayed immediately?
Bullets should be used sparingly and should contain short material, but they are useful to present taxonomies and high-level summaries. They also save time in preparation (vs. preferable visual slides). Bullets that are logical consequences of other bullets ... that follow from what was discussed ... should be hidden until that conclusion is natural. Bullets that answer questions should be hidden until the answer has been arrived at (interactively).
Is animation generally good or bad for lecture slides?
Animation for aesthetic purposes gives the feel of a board-room presentation in a marketing department. Animations should be as minimal as they can be ... for heavens sake, no whooshing around. However, animation of graphics can work well when explaining, for example, the dynamics of a system. It needs to be done well, however, and in clear stages! For example, here's a self-explanatory animation to teach bubblesort (from Wikipedia):
... it's faster and cleaner than going to the board, and one can even see the worst-case performance.
People who say "never use animation" (or make similar blanket statements) I feel lack imagination.
What is a good approach to take for titling slides?
Since the title of the slide is often the only substantial text on a slide, I use whatever title helps set a context for that slide in relation to others. I don't really believe in slide titles like "Sorting Algorithms I", "Sorting Algorithms II" ... instead ...
Should I include "separation slides" to chunk content?
Yes. I find separation slides (with a short large title) to be very useful to mark the end of a theme, to give a chance for pause and overcap, to emphasise the flow of the talk and to also give a brief opportunity for questions before moving on.
What is the optimal amount of information per slide?
That depends. Slides are generally free and content is easier to consume in sparse slides. Generally a picture or a diagram you can point to while talking is sufficient. But occasionally you might have a lot of content that is interdependent, such as for comparison:
(Taken from Wikipedia.)
This table is dense but it is best presented in it's entirety (with the help of colour) for the purposes of comparison. If the complexity of each algorithm were split over different slides, the comparison would be difficult to see.
A good rule of thumb is to include a minimal amount of content that stands as a cohesive whole: put the question and the answer on the same slide (since one makes little sense without the other), put the comparative data on one slide (since the comparison cannot be drawn otherwise), etc.
Slides should be numbered, but is it better to give the total number of slides on each (e.g., 4/20) or just the current slide (e.g., 4)?
I believe that parts should be numbered rather than slides. For me,
4/20 can sometimes be the psychological equivalent of making an audience watch a pot boil. Progress should instead be emphasised on the level of learning and topics (e.g., using separation slides) rather than an arbitrary slide numbering. "We've looked at X, now we're going to look at Y and Z. ... Okay, now that we've covered Y, all that's left for today is Z."
But slides should certainly be numbered! The audience/students may wish to ask a question or make a note about a specific slide!