Improve halfbyte transposition performance, marginally improving bbq performance

[benwtrent]

The transposition of the bits in half-byte queries for BBQ is pretty convoluted and slow. This commit greatly simplifies & improves performance for this small part of bbq queries and indexing.

Here are the results of a small JMH benchmark for this particular function.

TransposeBinBenchmark.transposeBinNew     1024  thrpt    5  857.779 ± 44.031  ops/ms
TransposeBinBenchmark.transposeBinOrig    1024  thrpt    5   94.950 ±  2.898  ops/ms

While this is a huge improvement for this small function, the impact at query and index time is only marginal. But, the code simplification itself is enough to warrant this change in my opinion.

[elasticsearchmachine]

Pinging @elastic/es-search-relevance (Team:Search Relevance)

[elasticsearchmachine]

Hi @benwtrent, I've created a changelog YAML for you.

[elasticsearchmachine]

💚 Backport successful

Status	Branch	Result
✅	8.17