Fix Arrow chunk download telemetry error bucketing - #1649

Description

An internal telemetry investigation found Arrow chunk download failures being emitted as DOWNLOAD_FAILED, which is an internal lifecycle state rather than a stable driver error code.

This change emits the existing CHUNK_DOWNLOAD_ERROR code for genuine download failures, preserves processing failures after download succeeds, and rethrows an existing typed DatabricksSQLException when a waiting chunk provider receives it. Untyped readiness failures still use CHUNK_READY_ERROR, and timeouts retain the original TimeoutException as their cause.

The change improves telemetry bucketing and exception preservation; it does not change download or retry behavior.

Testing

Focused tests cover successful download, body-read failure bucketing, malformed Arrow processing, typed future failures, untyped readiness failures, and readiness timeouts. The repository-wide Spotless check also passes.

Telemetry Errors

• Not applicable — this PR does not add or change a telemetry-visible error.
• Applicable — the error uses DatabricksDriverErrorCode where appropriate, and any
  new code is uniquely numbered and tested.
• Applicable — maintainer help is requested to confirm that the existing
  CHUNK_DOWNLOAD_ERROR and preserved processing-error classifications remain appropriate.

Additional Notes to the Reviewer

This intentionally reuses existing error codes; no new error enum is needed.