TurboQuant InnerProduct optimizations (#7396)

_Stacked on top of #7394

## Summary

Tracking issue: #7297

Implements the final pushdown / reduction rules needed to make the
TurboQuant quantization scheme make sense.

TODO

## Testing

More tests

---------

Signed-off-by: Connor Tsui <connor.tsui20@gmail.com>
Signed-off-by: Claude <noreply@anthropic.com>
Co-authored-by: Claude <noreply@anthropic.com>

Author: connortsui20

vortex-tensor/benches/similarity_search.rs

@@ -33,7 +33,7 @@ use common::generate_random_vectors;
 static GLOBAL: MiMalloc = MiMalloc;
 
 /// Number of vectors in the benchmark dataset.
-const NUM_ROWS: usize = 10_000;
+const NUM_ROWS: usize = 100_000;
 
 /// Dimensionality of each vector. Must be `>= vortex_tensor::encodings::turboquant::MIN_DIMENSION`
 /// (128) for the TurboQuant variant to work.