typed pattern slots; continuations; anchors; utt types

[jackft]

Purpose of this PR

• pattern slots now contain a "type" which signifies which kind of thing a pattern slot points to
• pattern slots can now point to timeline annotations
• "spans" are renamed to "textlets" for consistency
• there can be multiple 'utterance' tiers, e.g. speech and gesture per participant
  • this necessitated changing how token tiers are named
• we included anchors for aligning gesture and speech
• we included continuations, e.g. utterances which can be split over many lines but still point to one timeline annotation. The reason for this is to coordinate gesture & speech, for example, in the transcript without splitting up the underlying time alignment of a single gesture in the timeline.