"ANGULAR" distance defined differently than in main annoy package

[jrhaberstroh]

This Java implementation defines ANGULAR distance as "cosine similarity", while the C++/python implementation defines ANGULAR distance as "approximate theta".

Evidence:

Java implementation -- dot(u,v), the value of cosine

annoy-java/src/main/java/com/spotify/annoy/ANNIndex.java

Line 189 in aef275b

public static float cosineMargin(final float[] u, final float[] v) {

annoy-java/src/main/java/com/spotify/annoy/ANNIndex.java

Line 319 in aef275b

float margin = (INDEX_TYPE == IndexType.ANGULAR) ? cosineMargin(v, queryVector)

C++ / python implementation -- sqrt(2 * (1 - dot(u,v))), an approximation of theta
https://github.com/spotify/annoy/blob/6f6b0c84ab413337eb4d2e850a4cba637f52ccbc/src/annoylib.h#L473
https://github.com/spotify/annoy/blob/6f6b0c84ab413337eb4d2e850a4cba637f52ccbc/src/annoylib.h#L502

Related issue defining "angular" method in the other repo: spotify/annoy#530

Because this would be a significant breaking change, I did not put a PR together. Would the lead devs provide guidance on the appropriate way to name a new distance that implements angular correctly ("angular-cpp" or "angular-correct"?).