new feature: finer-granularity ErrorKind

### Feature Description

Hi team, when I'm using opendal to do write/read load test against internal S3, for `opendal_operation_errors_total` counter I'm getting all data point with `error` label as `Unexpected`.

I'm wondering if it's possible to have finer granular grouping? Having over 400 errors/second with all `Unexpected` errors make it hard to pinpoint and debug. :(

### Problem and Solution

Have finer granular error kind grouping.

### Additional Context

_No response_

### Are you willing to contribute to the development of this feature?

- [ ] Yes, I am willing to contribute to the development of this feature.

Start by tracing the opendal_operation_errors_total metric and ErrorKind references in the repository to understand how errors become metric labels. Define the finer-grained grouping and verify that load-test failures no longer collapse into Unexpected, with corresponding coverage for the revised labels.

Written by the indexing model from the issue text.