Improve markdown + HTML parsing

[nvlang]

This PR significantly improves SvelTeX's parsing of content mixing markdown and HTML syntax. Among other things, it includes the following changes:

• Use sanitize-html to ensure that HTML generated by Markdown
  processor is valid.

• Refine whitespace adjustment performed before passing markup to the
  Markdown processor.

• Remove <p> tags within HTML elements or Svelte components that
  cannot contain paragraphs (e.g., <span><p>*text*</p></span>
  becomes <span><em>text</em></span> now, ignoring insignificant
  whitespace.

• Add markdown.components option to SvelTeX configuration to specify
  preferences in regards to how each Svelte component is treated by
  SvelTeX when it comes to whitespace adjustments.

• Auto-import components "registered" in the markdown.components array
  from the SvelTeX configuration if they are used in the markup and not
  already imported in the file's <script> tag.

  Note: a component is "registered" in the markdown.components array
  iff there exists an object obj in the markdown.components array such
  that all of the following hold:

  • obj.name equals the name of the component (case-sensitive).
  • obj.importPath is not undefined.

• Add tests for all of the above features and fixes.

• Add markdown.remarkRehypeOptions and
  markdown.rehypeStringifyOptions to SvelTeX configuration when the
  unified Markdown backend is used.

Fixes #2.