1.  Create file `_includes/links.md` with definitions of all links.
2.  Replace links at the bottom of Jekyll'd `.md` files with inclusion of that file.
3.  Modify `bin/lesson_check.py` to load and check definitions in this file.

This eliminates most redundancy in link definitions, though
`CONTRIBUTING.md` and `README.md` still contain their own links, since
they are not Jekyll'd.  Note that we *cannot* put the links into
`_layouts/*.html` (or even `_layouts/something.md`) because of the
order in which Jekyll expands things - we have to explicitly include
it in each page.

1.  Created `bin/repo_check.py`.
2.  Moved `require()` to `util.py`.
3.  Updated `Makefile` with new target.

1.  Splitting commands to clean intermediate R files into separate target.  These files must be committed to the repo, so they shouldn't be deleted under normal circumstances.
2.  Modifying intermediate file generation to create fig/rmd-* rather than fig/swc-rmd-*, since this template is used by other carpentries (not just SWC).

1.  Renaming Ruby file to use underscore in name (consistent with other `bin` scripts).
2.  Generating `_includes/all_figures.html` for the first time when creating a new lesson.

1.  Moving YAML parsing to bin/util.py.
2.  Adding bin/extract_figures.py.
3.  Adding Makefile targets to regenerate image inclusion file.
4.  Adding _extras/figures.md to display all images.

on a suggestion of @sritchie73 and @aammd

Thanks to ocefpaf for request that.

Pandoc had some changes in the last versions.
Having a variable for Pandoc will help
maintainers to handle different versions of Pandoc.

This is based on 966a0322 from gh-pages.

Thanks to @ChristinaLK to report the problem.

First should come the common workflow tasks, then commands for internal/debugging use.