fix: restructuredText figure caption and alt

[trevorld]

• restructuredText figure caption is no longer escaped for html
• restructuredText figure caption and alt can now be independently set with fig.cap and fig.alt

• img.cap.rst() is a light fork of img.cap() in hooks-html.R that:

  1. Doesn't escape html if alt = TRUE
  2. If alt = FALSE won't try to add a figure "label" to the caption

fixes #2023

[trevorld]

I wonder if it's easier to add arguments to .img.cap() instead of forking it. Thanks!

One approach could be:

1. Add a new argument to .img.cap() with the target output format like output = "md" or output = "rst"
2. Update all calls to .img.cap() with this argument
3. Only do the html escaping and/or label creating for the formats that need/support them

• Do you want me to submit a revised PR that does this?
• I'm not entirely if the {knitr} codebase already has a standardized way to get a string of the target output format
• I'm not sure which formats really need an html escaped image alt or a label prefix prepended to the caption. Neither of those make sense in the restructuredText context. Presumably html output needs the html escaped image alt and latex output needs the label prefix?

[yihui]

I'll take care of the rest of work. Thank you!

• Presumably html output needs the html escaped image alt and latex output needs the label prefix?

Yes, that's about right. For rst output, the label won't be created.