[bug] Fix crash for completion edge case (#1668)

[scottnemes]

Description

Fixes crash on completion edge case described in #1668.

Underlying issue is that sqlparse has its own set of "keywords", and it appears that this means some columns that match like column_name in this case get parsed as the type Token vs the more specific type Identifier. I added logic to check this and in the case of a Token like column_name it will just return the value.

https://github.com/andialbrecht/sqlparse/blob/897eb2df8c0608b9ee4682e9728f669adf13595e/sqlparse/keywords.py#L175

Fixes #1668

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. Medium: Missing regression test for the fixed crash path

• Change at parseutils.py:262 adds a defensive type check in extract_columns_from_select, but this PR does not add a test that reproduces the Repeatable completions crash #1668 edge case and verifies completion no longer crashes.
• Action: add a regression test in completion-focused tests (for example test_smart_completion_public_schema_only.py) that exercises the problematic SELECT ... FROM input and asserts completion returns suggestions without exception.

No security issues were introduced in the diff.
No other clear correctness regressions were identified from the changed lines, but risk remains until the crash scenario is locked by test coverage.