…ch.md/autoresearch.sh/autoresearch.checks.sh infrastructure\n\nResult: {"status":"keep","chimp128_runtime_us":962.9880742729885}

…r-based fallback scan with manual bounded loop and use bitmask (&127) for modulo-128 indexing\n\nResult: {"status":"keep","chimp128_runtime_us":943.249552854506}

…emove unused decode lookup table writes\n\nResult: {"status":"keep","chimp128_runtime_us":934.797589779669}

…on/position lookups with direct match-based bin_count/bin_encode/bin_decode\n\nResult: {"status":"keep","chimp128_runtime_us":917.906839079959}

…value reference when hash lookup misses, removing per-sample 128-entry search\n\nResult: {"status":"keep","chimp128_runtime_us":759.7242453100638}

…eferencing, plus retain updated autoresearch notes\n\nResult: {"status":"keep","chimp128_runtime_us":749.2119429059167}

…lowest_bits/read_u64_lowest_bits and inline bit operations in src/bits.rs\n\nResult: {"status":"keep","chimp128_runtime_us":555.5697622385246}

…_and_code(xor) helper to compute bin count and 3-bit code in one match\n\nResult: {"status":"keep","chimp128_runtime_us":548.5704538310441}

… leading-zero count to 31 (5-bit field); checks now stable and chimp128 benchmark also improved\n\nResult: {"status":"keep","chimp128_runtime_us":545.0697426967323}

…rites and from_be_bytes slice read for count==64 when bit-aligned\n\nResult: {"status":"keep","chimp128_runtime_us":534.9991373464616}

…gle-byte fit fast paths for read_u64_lowest_bits/put_u64_lowest_bits (hot for chimp128 control fields)\n\nResult: {"status":"keep","chimp128_runtime_us":527.3073391516542}

…ncode to avoid reinitializing 16,384-entry lookup array each call\n\nResult: {"status":"keep","chimp128_runtime_us":526.951867330305}

…: return early with Vec::new for empty and use Vec::with_capacity(count) for decode results\n\nResult: {"status":"keep","chimp128_runtime_us":526.5303745140408}

…h direct 8-entry leading-zero table indexing on 3-bit codes\n\nResult: {"status":"keep","chimp128_runtime_us":526.0197709158932}

…ssing (unaligned head bits, branchless byte chunk loop, tail bits), cutting per-iteration branching overhead\n\nResult: {"status":"keep","chimp128_runtime_us":479.8389552574758}

… partial byte in one packed operation (both write and read) before byte-chunk loops\n\nResult: {"status":"keep","chimp128_runtime_us":447.33449102749296}

… remainder with single packed operation (write pending bitbuf directly, read from current byte with mask/shift)\n\nResult: {"status":"keep","chimp128_runtime_us":415.49565942946043}

…upEntry copy; read epoch/value directly from entries[key]\n\nResult: {"status":"keep","chimp128_runtime_us":402.32638972466026}

…est_bits/read_bit/read_u64_lowest_bits) after loop refactors; large win\n\nResult: {"status":"keep","chimp128_runtime_us":373.1848294725018}

…u64_lowest_bits/read_u64_lowest_bits (use put_u32/from_be_bytes before byte/tail handling)\n\nResult: {"status":"keep","chimp128_runtime_us":368.7577026392289}

…t_u64_lowest_bits, read_bit) on top of read_u64 inlining and chunked loops\n\nResult: {"status":"keep","chimp128_runtime_us":367.7745482436005}

…t_u64_lowest_bits plus 16-bit block reads in read_u64_lowest_bits after 32-bit chunks\n\nResult: {"status":"keep","chimp128_runtime_us":349.0826878012644}

…op) to reduce byte-loop iterations for mid-sized payloads\n\nResult: {"status":"keep","chimp128_runtime_us":344.0384787565408}

…generic for <=8 bits) and switch chimp128 hot 3/6/7-bit control fields to these paths\n\nResult: {"status":"keep","chimp128_runtime_us":334.3839392421396}

…ays (epochs + values) and access via single RefCell borrow in loop\n\nResult: {"status":"keep","chimp128_runtime_us":325.49438171286823}