feat: run comparison view for Scrape robots

[jceychen]

Hi @RohitR311 and @amhsirak,

This is @jceychen, and together with @Wootark-Kim and @Batkhuu831, we'd like to propose our next contribution.

Background
We noticed that Maxun already supports scheduled runs for robots, which is great for recurring tasks like tracking product listings, job postings, or news pages. We'd like to build on that by creating a notification that detects change(s) between consecutive runs, making it easier to identify updates without manually comparing outputs while providing a foundation that we or future developers can continue to expand on.

Proposed Feature (Initial Scope)
We want to add a run comparison feature to Scrape robots that:

• Adds a toggle in Robot Settings to turn on run comparison for that robot, so users can opt in on a per-robot basis
• Automatically compares the most recent run to the previous one when a scheduled run completes, and shows a notification if changes are detected
• Supports multiple output formats, starting with Text Content as the initial proof of concept

Future Possibilities

• Allows users to manually select any two runs from the runs view to compare directly, for cases outside of scheduled runs
• Clicking the notification to open a side-by-side view showing exactly what was added, removed, or unchanged between the two runs

Let us know what you think in terms of direction and/or scope before we get started!

[amhsirak]

@jceychen This looks great! 😄 Funny timing - I was actually thinking about building something very similar myself over the next week.

I think this is a solid direction. Go ahead with a small prototype that focuses on Text Content comparison only for now. Keeping the initial scope narrow will make it much easier to review and iterate on.

Once we have the prototype working well, we can stabilize and expand the feature in follow-up PRs.

[Batkhuu831]

Hi @amhsirak, I'm @Batkhuu831 , I worked on #1181 alongside with @jceychen and @Wootark-Kim on the Text Content comparison prototype.

With the prototype phase wrapping up, I'd like to lead the next phase as a solo contribution. You mentioned expanding the feature in follow-up PR's once the prototype stabilizes, I'd like to propose that next phase: extending comparison from Text Content to also cover Markdown and HTML.

Why these two:
Scrape robots already support Markdown and HTML as output formats, and both are stored in 'serializableOutput' the same way Text is, so this is comparison logic, not new storage. I'd leave Links, Summary, and Screenshots for later since they need different comparison strategies (structural or maybe an image diff) and would make the PR much bigger.

Proposed scope:

• Shared comparison logic used by both manual ('task-runner.ts') and scheduled ('scheduler/index.ts') runs, replacing the current duplicated Text-only logic
• Update 'GET /runs/:id/diff' to return previous and current output for Text, Markdown, and HTML
• Update the comparison dialog to let users pick a format when more than one is available
• Keep 'hasChanges' as a single boolean - true if any supported format changed

I know #1181 still has a few review comments open (the previous-run ordering logic, the missing 'hasChanges' in the run-list response). Happy to help get those resolved first if that's useful, and hold off starting this until #1181 is merged. Just wanted to float the direction and scope now so I can start as soon as it makes sense. Let me know what you think!