refactor test_cache_projection_excludes_nested_columns to use high level APIs

[alamb]

Which issue does this PR close?

• Related to Rewrite ParquetRecordBatchStream (async API) in terms of the PushDecoder #8677
• part of Rewrite ParquetRecordBatchStream in terms of the PushDecoder #8159

Rationale for this change

I am reworking how the parquet decoder's state machine works in #8159

One of the unit tests, test_cache_projection_excludes_nested_columns uses non-public APIs that I am changing

Rather than rewrite them into other non public APIs I think it would be better if this test is in terms of public APIs

What changes are included in this PR?

1. refactor test_cache_projection_excludes_nested_columns to use high level APIs

Are these changes tested?

They are run in CI

I also verified this test covers the intended functionality by commenting it out:

--- a/parquet/src/arrow/async_reader/mod.rs
+++ b/parquet/src/arrow/async_reader/mod.rs
@@ -724,7 +724,9 @@ where
             cache_projection.union(predicate.projection());
         }
         cache_projection.intersect(projection);
-        self.exclude_nested_columns_from_cache(&cache_projection)
+        // TEMP don't exclude nested columns
+        //self.exclude_nested_columns_from_cache(&cache_projection)
+        Some(cache_projection)
     }

     /// Exclude leaves belonging to roots that span multiple parquet leaves (i.e. nested columns)

And then running the test:

cargo test --all-features --test arrow_reader

And the test fails (as expected)

---- predicate_cache::test_cache_projection_excludes_nested_columns stdout ----

thread 'predicate_cache::test_cache_projection_excludes_nested_columns' panicked at parquet/tests/arrow_reader/predicate_cache.rs:244:9:
assertion `left == right` failed: Expected 0 records read from cache, but got 100
  left: 100
 right: 0
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace


failures:
    predicate_cache::test_cache_projection_excludes_nested_columns

test result: FAILED. 88 passed; 1 failed; 1 ignored; 0 measured; 0 filtered out; finished in 0.20s

Are there any user-facing changes?

No, this is only test changes

[alamb]

@XiangpengHao can I trouble you for a review of this PR (as you wrote the original version of the test)?

[XiangpengHao]

Looks good to me! This is much more readable than before 💯

[alamb]

Tahnk you @XiangpengHao