fix(Outreach): delete stale scraper_results.csv before each run

[octo-patch]

Fixes #256

Problem

Outreach.start() runs the scraper and then checks os.path.exists(output_path) to determine if scraping succeeded. However, .mp/scraper_results.csv always points to a fixed path (get_results_cache_path()). If the file already exists from a previous successful run, this check passes even when the current scraper invocation fails or times out — causing the outreach flow to send emails based on stale lead data.

Solution

Delete scraper_results.csv before invoking the scraper. After the scraper finishes, the existing os.path.exists() check is now reliable: the file can only exist if the current run produced it.

Testing

1. Run Outreach once so .mp/scraper_results.csv is created.
2. Cause the next scraper invocation to fail (e.g. kill the process or pass an invalid niche).
3. Before this fix: the stale CSV is found and outreach emails are sent to old leads.
4. After this fix: the stale CSV is removed before the run; when the scraper fails no new file is written; os.path.exists() returns False and the error path is taken correctly.