New APIs: ByteString.toIndex() and ByteString.toFraction()

[swankjesse]

The first one may be useful with hashing to put byte strings in
partitioning buckets for scaling. For example, to divide a dataset
into 32 partitions, hash the key then use toIndex(32) to map the
key to its partition.

The second one may be useful with dynamic experiments and A/B
tests. For example, to assign a control group to 5% of customers
hash the customer key then check if toFraction() is less than 0.05.

[swankjesse]

of everything, this part hurts me the most

[zhxnlai]

Great comments!

[JakeWharton]

Recording some out-of-band discussion points: we're looking for a word to replace "to" here since the index and fraction concepts are not intrinsic to domain of how you think of a bytestring. Jesse had a fun analogy where it's like having String.toIndex() which returned the absolute position of a string in a dictionary of all words. It's not perfect, of course, since strings can be arbitrary characters that aren't words. It is, however, a useful mental model since String.toIndex() would be a weird API but String.dictionaryIndex() makes sense. So what's the "dictionary" of the set of all possible bytes such that we can name these methods something like thingIndex() and thingFraction()?

[swankjesse]

I've been thinking about the names here. My biggest concern with the current names is they could be interpreted as decoding ASCII, like readDecimalLong.

My new recommendation is to drop the index function, and rename toFraction() to unitIntervalDouble(). You can get the result of toIndex by dividing that by the target size.