* `test_decoder_first_pages_only_batch_granularity`, the counterpart
  of `test_decoder_first_pages_only` from
  apache#11218. That test shows that the
  default `FetchGranularity::RowGroup` does not decode a batch until the
  whole row group is pushed. With `FetchGranularity::Batch`, pushing the
  first page of each column returns the first batch.
* `incremental_tests`: both granularities give the same batches for
  batch sizes around page and row-group sizes, projections with list and
  struct columns, page-skipping row selections with each selection
  policy, offset and limit, files without an offset index, predicate
  chains with and without the predicate cache (including cache misses),
  and 300 random combinations of all of these. Other tests check that
  requests serve one batch, that dictionary pages are requested once, that
  no range is requested twice, that bytes pushed ahead are released as the
  readers pass them, that the row-group boundary is visible after the
  last batch, and `into_builder` and `try_next_reader` interplay.

Co-Authored-By: Claude Opus 5.5 <[email protected]>

* `test_decoder_first_page_only_does_not_decode` from
  apache#11218, which shows that the
  row-group mode does not decode a batch until the whole row group is
  pushed, and a counterpart for `FetchGranularity::Batch` where pushing
  the first page of each column returns the first batch.
* `incremental_tests`: both granularities give the same batches for
  batch sizes around page and row-group sizes, projections with list and
  struct columns, page-skipping row selections with each selection
  policy, offset and limit, files without an offset index, predicate
  chains with and without the predicate cache (including cache misses),
  and 300 random combinations of all of these. Other tests check that
  requests serve one batch, that dictionary pages are requested once, that
  no range is requested twice, that bytes pushed ahead are released as the
  readers pass them, that the row-group boundary is visible after the
  last batch, and `into_builder` and `try_next_reader` interplay.

Co-Authored-By: Claude Opus 5.5 <[email protected]>

* `test_decoder_first_pages_only_batch_granularity`, the counterpart
  of `test_decoder_first_pages_only` from
  apache#11218. That test shows that the
  default `FetchGranularity::RowGroup` does not decode a batch until the
  whole row group is pushed. With `FetchGranularity::Batch`, pushing the
  first page of each column returns the first batch.
* `incremental_tests`: both granularities give the same batches for
  batch sizes around page and row-group sizes, projections with list and
  struct columns, page-skipping row selections with each selection
  policy, offset and limit, files without an offset index, predicate
  chains with and without the predicate cache (including cache misses),
  and 300 random combinations of all of these. Other tests check that
  requests serve one batch, that dictionary pages are requested once, that
  no range is requested twice, that bytes pushed ahead are released as the
  readers pass them, that the row-group boundary is visible after the
  last batch, and `into_builder` and `try_next_reader` interplay.

Co-Authored-By: Claude Opus 5.5 <[email protected]>

test_decoder_first_pages_only_batch_granularity is the counterpart of
the test from apache#11218. incremental_tests.rs checks that both
granularities give the same batches (targeted and randomized), and
checks the requests, the row-group boundary and the bytes held.

Co-Authored-By: Claude Opus 5.5 <[email protected]>

test_decoder_first_pages_only_batch_granularity is the counterpart of
the test from apache#11218. incremental_tests.rs checks that both
granularities give the same batches (targeted and randomized), and
checks the requests, the row-group boundary and the bytes held.

Co-Authored-By: Claude Opus 5.5 <[email protected]>

…ing a batch (apache#11218)

# Which issue does this PR close?

- Related to apache#6946

# Rationale for this change

While discussing apache#6946 with
@adriangb, a question came up about whether `ParquetPushDecoder` would
decode a batch if only the pages needed for that batch had been pushed
or if it insists on having all pages for the row group first.

This PR adds a test that answers that question

For the current code, the ParquetPushDecoder requires that all ranges
for the row group are present

# What changes are included in this PR?

 `test_decoder_first_page_only` test 

# Are these changes tested?

This PR is only a test.

# Are there any user-facing changes?

No.

---------

Co-authored-by: Claude Fable 5.1 <[email protected]>

test_decoder_first_pages_only_batch_granularity is the counterpart of
the test from apache#11218. incremental_tests.rs checks that both
granularities give the same batches (targeted and randomized), and
checks the requests, the row-group boundary and the bytes held.

Co-Authored-By: Claude Opus 5.5 <[email protected]>

test_decoder_first_pages_only_batch_granularity is the counterpart of
the test from apache#11218. incremental_tests.rs checks that both
granularities give the same batches (targeted and randomized), and
checks the requests, the row-group boundary and the bytes held.

Co-Authored-By: Claude Opus 5.5 <[email protected]>

test_decoder_first_pages_only_batch_granularity is the counterpart of
the test from apache#11218. incremental_tests.rs checks that both
granularities give the same batches (targeted and randomized), and
checks the requests, the row-group boundary and the bytes held.

Co-Authored-By: Claude Opus 5.5 <[email protected]>