Refactor shared execute() kwargs handling across cursor implementations

[laughingman7743]

Summary

Multiple cursor classes (sync/async + pandas/arrow/polars/s3fs) duplicate the same execute() argument handling for parameters like result_set_type_hints, result_reuse_enable, paramstyle, etc. Each new parameter added to execute() requires updating 11+ cursor files.

Proposal

Extract shared execute() kwargs into an ExecuteOptions dataclass or mixin to:

• Reduce duplication across cursor implementations
• Make it easier to add new execute() parameters in the future
• Centralize parameter validation and documentation

Affected files

• pyathena/cursor.py
• pyathena/async_cursor.py
• pyathena/aio/cursor.py
• pyathena/pandas/cursor.py, pyathena/pandas/async_cursor.py
• pyathena/arrow/cursor.py, pyathena/arrow/async_cursor.py
• pyathena/polars/cursor.py, pyathena/polars/async_cursor.py
• pyathena/s3fs/cursor.py, pyathena/s3fs/async_cursor.py

Definition of Done

• Adding a new execute() argument requires changes in 1-2 files (not 11+)
• sync/async + pandas/arrow/polars/s3fs cursors share the same validation logic (type checks, mutual exclusion, default resolution)
• execute() argument documentation lives in a single source (dataclass/mixin docstring) rather than being duplicated per cursor

Context

Identified during PR review of result_set_type_hints feature (#690). Scoped out to keep the initial PR focused on core functionality.