Resolves https://github.com/webrecorder/browsertrix/issues/2764
## Changes
Uses crawl first seed as starting URL instead of workflow first seed to
fix replay after saving workflow without running.
## Manual testing
1. Log in as crawler
2. Create a new workflow and crawl a single page, for example,
https://example.com/
3. Edit the workflow to change starting URL to https://example.org/
4. Save without running
5. Go to latest crawl tab, verify replay loads https://example.com/
04f5ec856f