Extract shared boilerplate from sync cursors into WithFetch mixin

[laughingman7743]

Problem

The five sync cursor implementations (Cursor, PandasCursor, ArrowCursor, PolarsCursor, S3FSCursor) duplicate ~350 lines of identical boilerplate code:

Duplicated code	Lines (×5 cursors)
Properties: result_set, query_id, rownumber, rowcount	~125
Fetch methods: fetchone, fetchmany, fetchall	~75
Lifecycle: close(), cancel(), executemany()	~60
execute() shared control flow	~100

This is the sync equivalent of the deduplication done for async cursors in #668 (WithAsyncFetch mixin).

Proposed solution

Create a WithFetch mixin in pyathena/common.py (mirroring WithAsyncFetch in pyathena/aio/common.py) that provides:

• Properties: arraysize (getter/setter), result_set (getter/setter), query_id (getter/setter), rownumber, rowcount
• Lifecycle: close(), executemany(), cancel()
• Default fetch: fetchone(), fetchmany(), fetchall() — delegates to self.result_set.fetch*()

Each specialized cursor would then only need to implement __init__, get_default_converter(), execute(), and format-specific methods (e.g., as_pandas(), as_arrow()).

Files involved

File	Change
pyathena/common.py	Add WithFetch(BaseCursor, CursorIterator, WithResultSet)
pyathena/cursor.py	Extend WithFetch, remove boilerplate
pyathena/pandas/cursor.py	Extend WithFetch, remove boilerplate
pyathena/arrow/cursor.py	Extend WithFetch, remove boilerplate
pyathena/polars/cursor.py	Extend WithFetch, remove boilerplate
pyathena/s3fs/cursor.py	Extend WithFetch, remove boilerplate

Context

• This was identified during Phase 3 async cursor work (Add native asyncio S3FS and Spark cursors with boilerplate deduplication (Phase 3) #668)
• The async side already has WithAsyncFetch in pyathena/aio/common.py as a working reference
• S3FSCursor will need async fetch overrides (same pattern as AioS3FSCursor)
• SparkCursor is excluded (different execution model, no result set)