Add bloom-filter with length

[mapleFU]

This patch add a data_index_bloom_encoding_stats.parquet with bloom-filter length. This file is generated using the arrow-rs:

cargo run --features=cli --bin parquet-rewrite -- -i data_index_bloom_encoding_stats.parquet -o data_index_bloom_encoding_with_length.parquet --compression none --bloom-filter-enabled true --bloom-filter-ndv 2000 

The statistics:

{
"Version": "1.0",
"CreatedBy": "parquet-rs version 49.0.0",
"TotalRows": "14",
"NumberOfRowGroups": "1",
"NumberOfRealColumns": "1",
"NumberOfColumns": "1",
"Columns": [
{ "Id": "0", "Name": "String", "PhysicalType": "BYTE_ARRAY", "ConvertedType": "UTF8", "LogicalType": {"Type": "String"} }
],
"RowGroups": [
{
"Id": "0", "TotalBytes": "199", "TotalCompressedBytes": "199", "Rows": "14",
"ColumnChunks": [
{"Id": "0", "Values": "14", "StatsSet": "True", "Stats": {" },
"Compression": "UNCOMPRESSED", "Encodings": "PLAIN RLE RLE_DICTIONARY ", "UncompressedSize": "199", "CompressedSize": "199", BloomFilter {"offset": "253", "length": "2064"}", OffsetIndex {"offset": "2342", "length": "11"}", ColumnIndex {"offset": "2317", "length": "25"}" }
]
}
]
}

The reason to re-generate a file is that, we still need to make sure the bloom-filter without length could be read.

[mapleFU]

@pitrou@wgtmac Would you mind take a look?

[pitrou]

This patch add a data_index_bloom_encoding_stats.parquet with bloom-filter length. This file is generated using the arrow-rs:

cargo run --features=cli --bin parquet-rewrite -- -i /data_index_bloom_encoding_stats.parquet -o data_index_bloom_encoding_with_length.parquet --compression gzip --bloom-filter-enabled true --bloom-filter-ndv 2000

Can you perhaps avoid using compression here?

[mapleFU]

Updated