Support `304` and non-modified content better

### Problem Description

This came up in https://github.com/elastic/crawler/pull/219

We currently don't support checking for modified content, but we could do this. If a server returns a `304` we just drop updating the content. Because `304` responses don't contain any content, we can't re-index the content from that URL as part of the response processing. This means that if a server has previously ingested a page when it returned `200`, but that page now returns `304`, the page will be deleted from the index and the content becomes invisible to Crawler.

### Proposed Solution

Crawler should treat `304` responses in the following way:

- Save URLs for `304` in a temporary list (in-memory)
- During purge crawl phase, if the URL is related to content from already indexed into ES, don't delete the doc
- Instead, update `last_crawled_at` value
- Output this information in debug logger

Start by tracing response processing for HTTP 304 responses and the purge crawl phase, then inspect how already-indexed Elasticsearch documents are selected for deletion. Verify the temporary URL list, preservation of related documents with an updated last_crawled_at value, and debug logging against the proposed behavior.

Written by the indexing model from the issue text.