[Refactoring] Generalize the ColumnReader and SelectiveColumnReader classes

[yingsu00]

This is a sub issue of #1533

ColumnReader, SelectiveColumnReader and SelectiveStructColumnReader classes need to be moved to facebook::velox::dwio::common::reader. However we need to note that some of their contents are just related to DWRF and ORC. For example:

• StripeStream and FlatMapContext are specific to DWRF

SelectiveColumnReader(
  	std::shared_ptr<const dwio::common::TypeWithId> requestedType,
  	StripeStreams& stripe,
  	common::ScanSpec* scanSpec,
  	const TypePtr& type,
  	FlatMapContext flatMapContext = FlatMapContext::nonFlatMapContext());

static std::unique_ptr<SelectiveColumnReader> build(
  	const std::shared_ptr<const dwio::common::TypeWithId>& requestedType,
  	const std::shared_ptr<const dwio::common::TypeWithId>& dataType,
  	StripeStreams& stripe,
  	common::ScanSpec* scanSpec,
  	FlatMapContext flatMapContext = FlatMapContext::nonFlatMapContext());

• RowIndex is DWRF specific

mutable std::unique_ptr<SeekableInputStream> indexStream_;
mutable std::unique_ptr<proto::RowIndex> index_;

• filterRowGroups means different in Parquet

std::vector<uint32_t> SelectiveColumnReader::filterRowGroups(
	uint64_t rowGroupSize,
	const StatsContext& context)

• ColumnStatistics means different in Parquet

	const auto& entry = index_->entry(i);
	auto columnStats =
        buildColumnStatisticsFromProto(entry.statistics(), context);
	if (!testFilter(filter, columnStats.get(), rowGroupSize, type_))

Here buildColumnStatisticsFromProto() is DWRF specific.
For Parquet, the columnstats are per RowGroup, and stored together in the footer. Filtering will be done in ParquetRowReader::filterRowGroups(). The file level ColumnStats from all ColumnChunk statistics need to be merged manually.

• null handling
  ColumnReader null handling doesn't apply to Parquet. Move out from

SelectiveColumnReader::prepareRead(
	vector_size_t offset,
	RowSet rows,
	const uint64_t* incomingNulls) 

Since We moved ColumnReader and SelectiveColumnReader, there are some leftover Dwrf specific leftovers and we may need to create a new level of ColumnReader super classes for Dwrf only. These will be DwrfColumnReader and DwrfSelectiveColumnReader in dwio::dwrf::reader. However If the leftovers are not too much we may also merge them in each Dwrf child XXXColumnReader classes.

[stale]

Is this still relevant? If so, what is blocking it? Is there anything you can do to help move it forward?

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs.