First of all, I'm being HEAVILY moderated in this forum. This reply is written on 6/18/2013 and not likely to hit the boards until 5 or 6 days later. I'd love to be proven wrong.
You've asked several questions here, but I'll focus on only two of them:
1. "Will update stats in parallel mode (with consumers) run as fast as my multi-command equivalent"? - Answer, I don't know. That would be worth testing, but keep in mind the requirements documented for "with consumers" clause well documented in the manuals. Mainly, this clause can only be used for update stats commands specifying columns, or columns lists. It's intended to be used on partitioned tables which live on RAID devices. So, there really isn't a comparison to be made here between this method (with consumers), and my method outlined in this thread.
2. "Does ASE break down an "update index statistics" command into it's basic parts and execute them "in parrallel"? - Answer - No. Not like I explained in this thread. It would execute each step (more or lesss) in a serial fashion, one after the other. For proof, examine the "lastmoddate" of sysstatistics, and systabstats for each column/partition/index/table of the indexes that have stats updated.
I suggest that you try to implement something similar to what I've suggested and let us know how that testing comes out. I can post something basic in terms of results for a test database that I have on a 2 - Engine linux box.