Enhance automated accessibility tests to check markup validity

[aduth]

🛠 Summary of changes

Updates accessibility feature specs to check for valid HTML markup.

Related Slack discussion: https://gsa-tts.slack.com/archives/C01C89LM6UF/p1673541645409329

The implementation ended up being more difficult than I expected, due to how the existing specs run using ChromeDriver, and Chrome will try to automatically repair much of the invalid HTML we're interested in.

📜 Testing Plan

1. Intentionally introduce errors to page HTML (e.g. git cherry-pick 226bee4)
2. rspec ./spec/features/accessibility
3. Observe failures

Example:

Expected page to have valid markup. Found syntax errors:
     
#<Nokogiri::XML::SyntaxError:"208:17: ERROR: That tag isn't allowed here  Currently open tags: html, body.\n  <p><div></div></p>\n                ^">
#<Nokogiri::XML::SyntaxError:"209:18: ERROR: Tag contains multiple attributes with the same name.\n  <div class=\"a\" class=\"b\"></div>\n                 ^">
#<Nokogiri::XML::SyntaxError:"211:1: ERROR: That tag isn't allowed here  Currently open tags: html, body, div.\n</body>\n^">

[nickttng]

Thank you for looking into this! 👍🏼

[zachmargolis]

LGTM