flash: Text file compression/decompression

[sampsyo]

There are already flash instructions for parsing text GFA and BED files (parse-gfa and parse-bed). It is common to distribute these files in compressed form. It would be nice to be able to parse these compressed files directly, without materializing the decompressed text file on disk.

Let's add new decompression instructions to flash. These could either fork the decompression tool as a subprocess or use Rust crates for in-process decompression, such as flate2, lzma-rust2, zstd, and so on.

[sampsyo]

In working on #272, I discovered a few deeper design problems to work on next:

• Figure out a way to avoid combinatorial code explosion when dealing with multiple instructions, multiple stream types, and multiple encodings.
• Figure out a way to force instruction implementations to either handle every encoding type or crash (but not silently proceed and ignore the encoding).

After that, we have the luxury of working on meatier functionality issues:

• We really need some sort of parallelism (flash: Parallel evaluation #271) if we're going to measure much useful here, because the baseline ("decompress to an OS pipe") will hang for any nontrivially-sized .gfa.gz file.
• Let's measure which flate2 backend is fastest.
• Support more compression schemes, especially zstandard.
• Compression (not just decompression). As in, writing compressed files.