[feat] Suggest related tables with foreign key relationships for JOIN and ON (#975)

[scottnemes]

Description

This PR adds foreign key relationship logic for completions:

1. After JOIN, tables with foreign key relationships to the given table will appear first in the list
2. After ON, the foreign key columns will be suggested to complete the JOIN

Example tables:

Note tables with foreign keys (orders and rosters) showing up first

Foreign key columns showing up after ON to join persons with orders

Same for rosters

Works with aliases as well

No foreign keys, just show the tables as expected

Resolves #975

Checklist

• I added this contribution to the changelog.md file.
• I added my name to the AUTHORS file (or it's already there).
• To lint and format the code, I ran

  uv run ruff check && uv run ruff format && uv run mypy --install-types .

[github-actions]

1. Correctness regression: FK ON suggestions can produce invalid SQL for quoted/reserved column names
   In sqlcompleter.py, table names are escaped, but FK column names are stored raw and later interpolated directly in _fk_join_conditions.
   Impact: for FK columns like order, user-id, or names requiring backticks, completion returns invalid expressions (e.g. o.order = u.id instead of o.`order` = u.id).
   Action: escape col/ref_col when persisting FK metadata (or when building condition strings), consistent with extend_columns() behavior.

2. Correctness edge case: JOIN prioritization can use FK metadata from the wrong schema
   In sqlcompleter.py, FK-priority lookup ignores parsed table schema and keys only by table name.
   Impact: for queries joining schema-qualified tables (e.g. other_db.orders), suggestions may be incorrectly prioritized using current-db FK metadata if table names overlap across DBs.
   Action: include schema in FK-priority matching, or explicitly skip FK-priority when parsed tables are from a non-current schema.

3. Missing tests for the above edge cases
   New FK tests in test_smart_completion_public_schema_only.py cover happy paths but not identifier escaping or cross-schema JOINs.
   Action: add tests for:

   • FK columns requiring escaping/backticks in ON condition suggestions.
   • JOIN prioritization when query uses schema-qualified tables outside current DB.

No direct security issues stood out in this PR diff.

[rolandwalker]

Legit! Going all the way with the full ON … = … is awesome.

In the case that the foreign keys exist, we should still allow completion on non-FK columns, just prioritize the FK case.