Hash Chain for Faster Match Finding

Naive longest-match search is O(window_size) per position — too slow for large windows. Real LZ77 encoders use a hash chain:

1. Build a hash of the next 3 bytes at each position: hash(text[i], text[i+1], text[i+2]).
2. Maintain a hash table: table[hash] = list_of_positions_with_that_3-byte-prefix.
3. To find a match at position p: hash the 3-byte prefix, look up the table, walk the list of candidates checking each.

The chain_limit is the speed/compression tradeoff knob. gzip's -1 to -9 levels mostly differ in this limit.

zstd uses a more sophisticated structure (suffix array / radix tree) for its higher levels. Snappy and lz4 use an even simpler one-position-per-hash table — fast but lower compression.