Experimental Configuration Clarification

[MGibson1]

I've been looking into the differences between the WhatsAppV1Configuration and ExperimentalConfiguration<L>. Beyond the obvious domain binding, I'm kind of struggling to understand the reasons behind differences I see. From what I surmise, these appear to be the meaningful changes:

1. Empty values are handled different (hashed zeroes in whatsapp and zeroes in experimental)
   • WhatsApp: hashed zero-values
   • Experimental: Zero-valued arrays
2. commitment nonce's are bound differently
   • Whatsapp: Key, label, version, value
   • Experimental: Key, Label, namespace
3. The empty label has changed
   • WhatsApp: [1u8;32]
   • Experimental: [1u8, 0u8 x 31]

Here are my best guesses for the reasoning behind changes

1. Performance and visibility. Reduces runtime hashing of constant values and human-recognizable empties
2. It's not clear to me why this was done. Unless there's some property that makes consistent label nonces desirable, this seems like a bad idea. I base that only on the name nonce, though. I haven't yet figured out why this needs to be a nonce.
3. I don't see a benefit to this one.