### What problem does this PR solve?

Issue Number: None

Related PR: apache#69159

Problem Summary: The raw SEARCH score suite requires expression pushdown and inverted-index execution. Fuzzy sessions can disable expression pushdown, leaving SEARCH to row evaluation and causing an internal error. Set both options explicitly before the suite runs.

### Release note

None

### Check List (For Author)

- Test: The identical suite file passed CloudP0 in apache#69160 (build 1072387); git diff --check passed. CI for this PR is pending.
- Behavior changed: No. Test setup only.
- Does this need documentation: No.

### What problem does this PR solve?

Issue Number: DORIS-29489

Problem Summary:

`SEARCH` with `score()` over a VARIANT sub-column fails on this branch with

```
ERROR 1105 (HY000): errCode = 2, detailMessage = Index statistics collection failed:
Not such column 2.v.content
```

`SearchPredicateCollector::collect_from_leaf` resolved the clause's field name only through
`tablet_schema->field_index(field_name)`. A VARIANT path such as `v.content` is not a
tablet-schema column while the path is not materialized, so the clause returned early and the
collection statistics were gathered for no field at all. The read side then asks the statistics
for the same path (`<parent uid>.<parent>.<path>`, the Lucene field the index files were written
with) and throws `INVERTED_INDEX_CLUCENE_ERROR` as `Not such column`.

`MatchPredicateCollector` already resolves the same shapes from schema: materialized/typed
paths, `field_pattern` templates, and a plain parent index inherited by the dynamic VARIANT
placeholder. This PR moves that resolution into one helper
(`resolve_scoring_index_candidates`) and uses it from the SEARCH collector too. When the clause
field is not a tablet-schema column, the collector now builds the same synthetic VARIANT column
the scan uses for a dynamic path (parent unique id + full path) and resolves it through the
parent index, keeping the cloned index metadata alive through `CollectInfo::owned_index_meta`.
Unresolvable names keep the previous skip, so only VARIANT paths gain collection.

The helper also carries the leaf type of the resolved shape, so the reader-selection and BKD
rejection this branch gained from the apache#69051 backport (apache#69159) keep working for plain fields and
for an inherited VARIANT parent index, which must be scored as text rather than as a BKD leaf.

### Release note

`score()` now works over `SEARCH` on VARIANT sub-columns served by a parent VARIANT inverted
index.

### Check List (For Author)

- Test <!-- At least one of them must be included. -->
    - [x] Unit Test
      `CollectionStatisticsTest.SearchCollectsVariantParentIndexStatistics` covers the collector
      resolution; run through `run-be-ut.sh`.
    - [x] Regression test
      `regression-test/suites/search/test_search_variant_subcolumn_bm25.groovy` runs the reported
      query plus the phrase and `field_pattern` variants.
    - [ ] Manual test (add detailed scripts or steps below)
    - [ ] No need to test or manual test. Explain why below to give a good reason
- Behavior changed: Yes (SEARCH + score() on VARIANT sub-columns returns scored rows instead of
  failing with `Not such column`)
- Does this need documentation: No