ARROW-9177: Add Hadoop-produced LZ4-compressed file with several frames
It seems than when the decompressed size exceeds 128 kiB, Hadoop compresses the data in several concatenated "frames".
Data in this file:
```
Version: 1.0
Created By: parquet-mr version 1.11.1 (build 765bd5cd7fdef2af1cecd0755000694b992bfadd)
Total rows: 10000
Number of RowGroups: 1
Number of Real Columns: 1
Number of Columns: 1
Number of Selected Columns: 1
Column 0: a (BYTE_ARRAY/UTF8)
--- Row Group: 0 ---
--- Total Bytes: 400029 ---
--- Rows: 10000 ---
Column 0
Values: 10000, Null Values: 0, Distinct Values: 0
Max: ffffe6a0-e0c0-4e65-a9d4-f7f4c176aea2, Min: 00087de7-10df-4979-94cf-79279f9745ce
Compression: LZ4_HADOOP, Encodings: BIT_PACKED PLAIN
Uncompressed Size: 400029, Compressed Size: 358351
--- Values ---
a |
[ ... ]
```
1 file changed
