[WIP] Add merchant/entity resolution engine package

[Copilot]

Thanks for assigning this issue to me. I'm starting to work on it and will keep this PR's description up to date as I form a plan and make progress.

Original prompt

---

This section details on the original issue you should resolve

<issue_title>feat(resolution): merchant/entity resolution engine — the core differentiator</issue_title>
<issue_description>## Summary
Build packages/resolution — the most important package. This is what makes FinancialAdvisor actually useful. This package does not currently exist in meaningful form.

Owns

• Merchant/entity resolution: "AMZN MKTP US*2K7X9" → "Amazon" with confidence score
• Alias graph: Maintain a graph of merchant name variants → canonical entity
• Duplicate clustering: Detect duplicate transactions across imports/accounts
• Transfer detection: Identify internal transfers (matching opposite amounts, close dates)
• Recurring series detection: Find subscriptions, bills, salary (periodic amount patterns)
• Explanation/confidence generation: Every resolution produces a confidence score + reasoning
• User correction learning: When user corrects a resolution, learn from it

Resolution Pipeline

Raw description → Text normalization (strip noise, uppercase, trim) →
Fuzzy match against alias graph → Confidence scoring →
If low confidence: flag for review → User decision → Update alias graph

Fuzzy Matching Strategy

1. Exact match against alias graph
2. Levenshtein/Jaro-Winkler distance matching
3. Token overlap ("AMAZON" in "AMZN MKTP US*2K7X9")
4. Historical pattern matching (this account + similar amount = likely same merchant)
5. Optional: embedding-based semantic similarity (ai-providers, not required)

Transfer Detection

• Match transactions across accounts: opposite amounts, within 3-day window
• Flag for review, don't auto-resolve (false positives are costly)

Recurring Detection

• Group by merchant → analyze date intervals → detect periodicity
• Handle amount variance (utility bills fluctuate)
• Confidence: strict periodic = high, irregular = low

Acceptance

• Merchant resolution handles common bank description formats
• Alias graph persists and grows from user corrections
• Transfer detection finds obvious internal transfers
• Recurring detection finds monthly subscriptions
• Every resolution includes confidence score</issue_description>

Comments on the Issue (you are @copilot in this section)

• Fixes feat(resolution): merchant/entity resolution engine — the core differentiator #53

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.

[kayodebristol]

Closing this PR to enforce the 1-PR-per-repo rule. Only one Copilot PR is allowed per repository at a time. The oldest PR (#48) will be kept active.