fix: validate AI response and guard downloadPDF against empty data

[manya-singh7]

first commit belongs to PR #37

What changed

Fixes #10

Note: The first commit in this PR belongs to PR #37 (already open).
The actual fix is in the second commit.

How to test

1. Enter a URL that blocks scrapers (e.g. a JavaScript-heavy site with no static text)
2. Click Generate and wait through all 4 stages
3. Observe: a clear error message appears instead of a blank PDF downloading

Visual Changes

No direct UI changes — this is a logic/validation fix.
The visible result is an inline error message now appears instead of
silently generating a blank resume.

Quality Checklist

Testing Locally

• The app still loads and works end-to-end (paste a URL, generate, download).
• Your change works on mobile (Chrome DevTools → responsive mode).
• [] Your change works in Chrome, Firefox, and Safari (if relevant).
• No new console errors or warnings are generated.
• [] If you changed PDF output, the generated PDF looks correct.
• [] If you changed the UI, it still matches the overall dark aesthetic.

Code Quality

• My branch is completely up to date with main.
• No commented-out debug code has been left in.
• CSS variables are used for all colors (no hardcoded hex values that break theming).
• All changes remain within index.html (unless strong architectural reasons were approved).

GSSoC '26 Compliance

• This issue was officially assigned to me before opening this PR.
• I have labeled this PR with gssoc26 in the description.
• This communication and my code adjustments are completely my own work.

[vercel]

@manya-singh7 is attempting to deploy a commit to the Abhijith's projects Team on Vercel.

A member of the Team first needs to authorize it.