[GLUTEN-10151][CELEBORN] Bump Celeborn version to 0.6.0

[SteNicholas]

What changes were proposed in this pull request?

Bump Celeborn version to 0.6.0. Backport:

• [CELEBORN-1577][Phase1] Storage quota should support interrupt shuffle. celeborn#2801
• [CELEBORN-1645] Introduce ShuffleFallbackPolicy to support custom implementation of shuffle fallback policy for CelebornShuffleFallbackPolicyRunner celeborn#2807
• [CELEBORN-1648] Refine AppUniqueId with UUID suffix celeborn#2810
• [CELEBORN-1896] delete data from failed to fetch shuffles celeborn#3109
• [CELEBORN-1856]Support stage-rerun when read partition by chunkOffsets when enable optimize skew partition read celeborn#3118
• [CELEBORN-1719][FOLLOWUP] Rename throwsFetchFailure to stageRerunEnabled celeborn#3324
• [CELEBORN-1921] Broadcast large GetReducerFileGroupResponse to prevent Spark driver network exhausted celeborn#3158

Fixes: #10151

How was this patch tested?

CI.

[github-actions]

#10151

[github-actions]

Run Gluten Clickhouse CI on x86

[zhouyuan]

@SteNicholas It seems the C++ client is ready, so does this mean actually we can use the c++ client in native shuffle?
Support CppClient in Celeborn
https://celeborn.apache.org/community/release_notes/release_note_0.6.0/

[SteNicholas]

@zhouyuan, you could use the experimental C++ client of celeborn for native shuffle.

[SteNicholas]

Ping @kerwin-zk.

[github-actions]

Run Gluten Clickhouse CI on x86

[zhouyuan]

@kerwin-zk @SteNicholas Thanks. Do you happen to have some reference perf numbers on C++ client vs Java client?

[kerwin-zk]

@kerwin-zk @SteNicholas Thanks. Do you happen to have some reference perf numbers on C++ client vs Java client?

@zhouyuan Currently, there is only approximate data, roughly about a 10% improvement.