add DSLX sample for Dictionary Based Encoding (DBE) encoder/decoder

[proppy]

See https://en.wikipedia.org/wiki/Dictionary_coder

This will help demonstrate how to implement streaming algorithms that heavily rely on stateful memory with DSLX procs.

[proppy]

Possible algorithms researched:

• Adaptive Lossless Data Compression Algorithm (ALDC) w/ known hardware implementations (1, 2)
• https://github.com/google/brotli
• https://facebook.github.io/zstd/

[rdob-ant]

In addition to the above, we're also considering two other algorithms:

• DEFLATE - very commonly used (e.g. in .zip / .gz file formats), based on LZ77-like dictionary coder used together with Huffman encoding.
• LZ4 - based on pure LZ77-like dictionary coder. Commonly used where especially fast decompression is required (that comes at the expense of reduced compression ratio compared to DEFLATE/Brotli/Zstd).

For me, it is tempting to go with LZ4. Due to its relatively widespread use it would make a much more "useful" DSLX sample, while still being similar to the initially considered ALDC (so most of our ALDC-related ideas are still applicable), and at the same time not resulting in much scope creep (e.g. adding Huffman or entropy coders/decoders on top of the dictionary coder - which would be required for most other algorithms).

[proppy]

Also learned about the following today:

• https://sneller.io/blog/decompressing-at-over-10-gigabytes-per-second/
• https://github.com/inikep/lizard
• https://github.com/SnellerInc/sneller/tree/master/ion/zion/iguana

[rdob-ant]

Opened draft #1038 which shows current state of matters