drop renaming of zone id

[mkreilly]

this crashes and @conorhenley said he has been commenting it out. @fscottfoti can we drop this line because the problem in orca has now been fixed? and all the comments above it? or?

[fscottfoti]

OK, it took a little thinking but I think this is what is going on.

zone_id_x probably doesn't exist because of this commit UDST/orca@614b346 I made in April of 2017. Basically we make sure that when two tables are joined that have the same column, one of them keeps the original column name instead of being _x and _y. Thus I think the line can be removed.

I do still think the core issue is lurking having to do with the cache, as this issue is still open and the default caching hasn't changed. UDST/orca#16 I think it's probably ok for reasons too complex to go into here, but I'd be wary.

[mkreilly]

Thanks for pondering this! I will merge in the change but leave it as commented out with the comments above as a reminder that there is this issue lurking in the code :(

[theocharides]

@mkreilly I'm just now catching up with the issues/comments on this. SO, commenting that line out is what made our jobs totals low during Horizon. The code returns back to its old behavior with null zone_ids, because we're not using zone_id_x which is unaffected by the cache issue. When you add the line back in, the problem was fixed on my machine, however it crashed when trying to run on the server (as @conorhenley was experiencing).

The crash is due to the update to merge_tables that @fscottfoti mentions. The server must have an updated version of orca, so it didn't work. I've had to add lines that force zone_id_x in so merge_tables can't get rid of it, in order to return to using it as a way of circumventing the cache issue. That is still an open issue: UDST/orca#16