CSV import: match columns case-insensitively (with opt-out for case-sensitive mode)

CSV import currently does case-sensitive column matching. If an imported CSV has `EmailAddress` and the existing table has `emailAddress` (or `email_address`), the import treats it as a new column instead of matching the existing one. This is friction for users importing CSVs from external sources that don't match Harper's exact column casing.

## Ask

Detect case-only (and possibly camel↔snake) differences during CSV import, and match against the existing column rather than creating a new one.

## Design considerations

- **Default behavior:** case-insensitive match. Optional flag (e.g. `caseSensitiveColumns: true`) to opt out.
- **Ambiguity:** if a CSV has both `name` and `Name` columns AND the table has only `name`, what wins? Reject the import with a clear error rather than picking arbitrarily.
- **Snake_case ↔ camelCase:** out-of-scope for this ticket unless trivial — opening a second ticket if asked.

## Acceptance criteria

- Case-only mismatch between CSV column and existing attribute uses the existing attribute.
- Ambiguous cases (two CSV columns map to the same attribute) are rejected with a clear error.
- Tests cover: exact match, case-only mismatch, ambiguous mismatch.

---

🤖 Filed by Claude on behalf of Kris.