Thanks for the feedback!
The delay feature is alive and well and has been expanded in the latest version (see mosaic button that allow you to split the capture screen and get multiple streams from the same camera at different delay values).
What has changed is that when recording video, we previously recorded whatever was displayed in the capture screen, so if you had a 10-second delay and recorded 5 seconds of video, at the end of the recording you have on file an action that happened before you hit the record button.
The new behavior (since 0.8.24, March 2015) is to record the stream as it comes from the camera, independently of what is shown on screen. So if you have a 10-second delay and record 5 seconds of video, at the end of the recording you have the action following your hitting the record button.
This was done to improve the recording performance and avoid dropped frames. Dropped frames cause mismatches between real time and video framerate and break time and speed measurements when analyzing the videos, so it's very important to get this right.
I understand that there is now a missing feature of being able to record action in "pre-roll" fashion and considering various options.
Another feature that is often requested is to be able to increase the size of the delay buffer. This is currently limited by memory. It should be possible to increase it on 64-bit versions, but the real solution would be to allow the delay buffer to be stored on disk. I feel these two features are interlinked because if we have the delay buffer in a temporary file on disk, it should be a small step to being able to save the part of the delay buffer between the start and stop record commands to be converted to a regular video file.