Add more situations to the angle bracket test.

[adam-azarchs]

Examples to demonstrate
#137.

Currently these tests will fail. I actually don't know where to find the code behind util.htmlEscape, which is what would need to be fixed.

[adam-azarchs]

The test failure is

FAIL: Actual did not match golden.
11c11
< in a `&lt;code block&gt;`.
---
> in a `<code block>`.
70c70
< foo = "&lt;thing&gt;"
---
> foo = "<thing>"
80c80
< | <a id="bracket_function-md_string"></a>md_string |  A markdown string.   |  <code>"`1&lt;&lt;10`"</code> |
---
> | <a id="bracket_function-md_string"></a>md_string |  A markdown string.   |  <code>"&#96;1&lt;&lt;10&#96;"</code> |
88c88
< deprecated for &lt;reasons&gt; as well as `&lt;reasons&gt;`.
---
> deprecated for &lt;reasons&gt; as well as `<reasons>`.
103c103
< foo = "&lt;brackets&gt;"
---
> foo = "<brackets>"

which are demonstrative of the problems described in #137.

In some of these, the problem could be resolved by replacing ` with <code>, though I'd consider that sub-optimal because it makes the markdown less idiomatic. It also doesn't really work for the code block (```) case - while that could be replaced by <pre> blocks, the result wouldn't get syntax highlighting, which would be unfortunate.

The md_string argument case is maybe enough of an edge case to not worry about it. I don't like that idea much on principle, though.