[mcp] New tool: deliveriq_email_pattern_finder

[Davison-Francis]

Given a domain (e.g. acme.com), infer the most likely email pattern by probing 6-8 known seed names — john.smith@, jsmith@, john@, smith.j@, etc. — and returning the pattern with a confidence score.

Why: today the user has to call deliveriq_find_email once per known person and have the model pattern-match manually. A dedicated tool is ~5× faster and more reliable.

Inputs: domain, optional list of known sample names.
Outputs: pattern string ({first}.{last}@{domain}), confidence (0-1), evidence (which seeds matched).

Credits: 6 (one verify call per seed).

Happy to take a PR.

[Davison-Francis]

Targeted for v1.2.0.

The functionality already exists server-side via two endpoints:

• /api/v1/email-finder/find (in deliveriq_find_email) — finds an email by name + domain using pattern recognition (8 common patterns: first.last@, firstinitial.last@, first@, etc.).
• /api/v1/org-intel/{domain}/patterns — returns the most common email pattern for a domain (e.g. {first}.{last}@example.com).

The second endpoint isn't yet exposed as an MCP tool — that's the gap this issue is calling out. Will add deliveriq_email_pattern (returning the dominant pattern + confidence + sample size) in 1.2.0.

Naming: would prefer deliveriq_email_pattern (singular, since it returns the canonical pattern, not all variants). Open to feedback if deliveriq_email_pattern_finder is clearer.