Function quantize_vec_with_codebook 

pub fn quantize_vec_with_codebook(
    config: &TurboQuantConfig,
    data: &[f32],
    codebook: &Codebook,
    sign_pattern: &[f32],
) -> Result<PackedBlock>

Quantizes a floating-point vector into a packed PackedBlock using pre-fetched codebook and sign pattern.

Hot-path variant of quantize_vec: avoids repeated codebook allocation and sign-pattern generation when quantizing many vectors with the same config (e.g. in batch KV-cache insertion during prefill).

Integration: orchestrates check_values_match, l2_norm, normalize_inplace, rotate, quantize_coordinates, select_scale, and PackedBlock::new – all using caller-provided codebook and sign pattern.

Errors

Returns [TurboQuantError::DimensionMismatch] if data.len() != config.dim.