As I've been involved in the recording of such material, perhaps I can explain...

What tends to happen is the items are performed two or three times prior to the concert, so that close-up shots can be taken without the cameras intruding too much into the concert itself.

The audio track is one 'genuine' take (normally the concert itself), but the video could be from any of the other 'takes', which means the video editor has to try to match the video with the audio. How well they do it is often more to do with how consistent the performer is rather than the skill of the editor.

So it may appear to be 'lip-sync' when really it isn't. I can't comment on the Celtic Woman concert, as I wasn't involved, but maybe this answers your question?