fix: plain domain entries should match subdomains

[vparfonov]

Summary

• Plain domain entries like example.com now match the domain itself and all subdomains (e.g. sub.example.com), aligning with Go's httpproxy, curl, Python, and Ruby conventions
• The leading-dot form .example.com continues to match subdomains only (not the domain itself), preserving a useful distinction
• Updated existing tests and added new plain_domain_matches_subdomains test

Problem

NoProxyItem::Plain performed exact string comparison only:
Self::Plain(source) => source == value, this meant no_proxy=apps.example.com would not bypass the proxy for elasticsearch.apps.example.com, contradicting all major NO_PROXY implementations and the GitLab standardization
proposal which states: "Suffixes are always matched (e.g. example.com will match test.example.com)"
https://about.gitlab.com/blog/we-need-to-talk-no-proxy/

Fix

Self::Plain(source) => {
      source == value || value.ends_with(&format!(".{source}"))
  }

Should fix: #11

[vparfonov]

@jdrouet can you please take a look