#1816: review use of <caption>

Has this been used consistently? Is it important to distinguish these blocks of texts as "captions" vs anything else.

Rendering behavior as of this post defaults to below figure and centered. This makes actual rendering to match facsimile more difficult, as seen on 14r and 94v.

14r:

Here is an example where the letters labeling the figures are not transcribed or encoded as captions (19v):