[GLUTEN-6235][CH] Fix crash on ExpandTransform::work()

[exmy]

What changes were proposed in this pull request?

The pr #5254 would filter redundant attribute when transforming to substraint plan in order to fix Unexpected empty column exception when reading hive text table. This introduces a new issue, for example:

(1) NativeScan hive tmp.testxmy_expand
Output [11]: [a1#4, a2#5, a3#6, a4#7, a5#8, a6#9, a7#10, a8#11, a3#6, a4#7, a9#12]
Arguments: [a1#4, a2#5, a3#6, a4#7, a5#8, a6#9, a7#10, a8#11, a3#6, a4#7, a9#12], HiveTableRelation [`tmp`.`testxmy_expand`, org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe, Data Cols: [a1#4, a2#5, a3#6, a4#7, a5#8, a6#9, a7#10, a8#11, a9#12], Partition Cols: []]

(2) ExpandExecTransformer
Input [11]: [a1#4, a2#5, a3#6, a4#7, a5#8, a6#9, a7#10, a8#11, a3#6, a4#7, a9#12]
Arguments: [[a1#4, a2#5, a3#6, a4#7, a5#8, a6#9, a7#10, a8#11, a3#6, a4#7, a9#12, 0], ....

After filter redundant attribute, the HiveTableScan operator output 7 distinct attributes in substrait read node but its output method still output 9 attributes include 2 duplicated arrtibutes, which causes column index mismatch in following process and crash.

We should keep the HiveTableScan's output attributes unique.

(Fixes: #6235)

How was this patch tested?

Add UT

[github-actions]

#6235

[github-actions]

Run Gluten Clickhouse CI

[github-actions]

Run Gluten Clickhouse CI

[zzcclp]

LGTM