Add clustering binding

[pedramr]

Running the example:

Clustering 10000 vectors of dimension 64 into 10 clusters

Running simple k-means clustering...
Simple k-means completed in 8.759083ms
Average quantization error: 2698.03

Running clustering with custom parameters...
Sampling a subset of 2560 / 10000 for training
Clustering 2560 points in 64D to 10 clusters, redo 3 times, 25 iterations
  Preprocessing in 0.00 s
Outer iteration 0 / 3
  Iteration 24 (0.01 s, search 0.00 s): objective=2.69803e+07 imbalance=1.296 nsplit=0
Objective improved: keep new clusters
Outer iteration 1 / 3
  Iteration 24 (0.01 s, search 0.01 s): objective=3.06187e+07 imbalance=1.357 nsplit=0
Outer iteration 2 / 3
  Iteration 24 (0.02 s, search 0.01 s): objective=2.0157e+07 imbalance=1.134 nsplit=0
Objective improved: keep new clusters

Advanced clustering completed in 19.593958ms

Comparing clustering quality...
Average distance to nearest centroid:
  Simple k-means:   10705.22
  Advanced method:  8305.35
  Improvement:      22.42% better

Time comparison:
  Simple:   8.759083ms
  Advanced: 19.593958ms (2.2x slower due to multiple runs)