Fix variable nanosecond timestamp update_time from response

[chemelnucfin]

@tswast Thanks for the hint.

Closes#4807

[tswast]

LGTM, though we might need a fallback in case microseconds appear again.

We should check with the product team to see what they are committing to.

[chemelnucfin]

Would a try except work better then or should we just wait?

try:
return ms
except ValueError:
return ns

[tswast]

I'm emailing the team to check.

[tseaver]

@tswast ping?

[tswast]

I never got a response from the folks I've contacted before on the RuntimeConfig team.

@mgannholm Can you comment on this?

[tswast]

Okay, I got confirmation from Victor on the RuntimeConfig team that "it might change from platform to platform". We need to be able to handle both.

Yes, try-except would be a good way to do it. Alternatively, I believe we have places (in BigQuery, maybe?) where we try to parse the string a little bit to see how many digits we have before trying to parse.

[chemelnucfin]

I'll wait until the datetime PRs #4979, #4980, #4981 are merged before finishing this up.

[chemelnucfin]

@tswast PTAL at your convenience.

[chemelnucfin]

going to merge this and diff from this in the new PR. (I have it written already)