Create a more flexible HTML comment parser and raise warning

[yluobb]

Towards: #501

This PR builds on top of the work from here. Create a more flexible HTML comment parser so we don't rely on spaces to exist to succesfully parse HTML comments. Also, raise warnings when there are missing spaces.

[linux-foundation-easycla]

• ✅login: yluobb / (70fa4d2)
• ✅login: jsuereth / (230a23e)
• ✅login: jsuereth / (230a23e, b8d7626)
• ✅login: jsuereth / (230a23e, b8d7626, cbc790e)

The committers listed above are authorized under a signed CLA.

[jsuereth]

Three things:

1. I don't think we need to force spaces here.

2. For the purpose of separation of concerns, the validation logic should go in parse_semconv_snippet_directive. Ideally this function JUST parses HTML headers, which have no restrictions on whitespaces.

3. We're not using eprintln! for warnings, instead you'll want to find a way to return a Diagnostic, e.g. using weaver's result WResult. This might require some wiring to make sure non-fatal-errors (NFE) can make it out of the parser. I haven't tried to sort out NFE w/ the nom parsers bits yet. Let me know if you have issues I'll see if I can put a scaffold up for you there.

[jsuereth]

A quick thought on warnings here:

1. You can update the update_markdown_contents function to use weaver result.
2. You can update the logic to something like:

  if parser::is_markdown_snippet_directive(line) {
    ... existing ...
  } else if parser::is_possible_markdown_snippet_directive(line) {
     ... issue warnings ...
 }

Where is_possible_markdown_snippet_directive is a function you can write re-using the HTML comment parser with some kind of very flexible mispelling finding parser.