Runtimeconfig: Update time fails to parse

[kw]

OS: Linux (Debian 4.9.65)
Python: 3.5.3 / 2.7.13
Version: google-cloud-runtimeconfig==0.28.0, google-cloud==0.32.0

Steps to reproduce:

1. Using gcloud, set a runtimeconfig variable.
2. In Python, load that variable
3. Try to access its update_time property.

>>> from google.cloud import runtimeconfig
>>> client = runtimeconfig.Client()
>>> config = client.config('testconfig')
>>> var = config.get_variable('foo')
>>> var.update_time
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/redacted/lib/python3.5/site-packages/google/cloud/runtimeconfig/variable.py", line 160, in update_time
    value = _rfc3339_to_datetime(value)
  File "/redacted/lib/python3.5/site-packages/google/cloud/_helpers.py", line 274, in _rfc3339_to_datetime
    dt_str, _RFC3339_MICROS).replace(tzinfo=UTC)
  File "/usr/lib/python3.5/_strptime.py", line 510, in _strptime_datetime
    tt, fraction = _strptime(data_string, format)
  File "/usr/lib/python3.5/_strptime.py", line 343, in _strptime
    (data_string, format))
ValueError: time data '2018-01-22T21:39:44.095040522Z' does not match format '%Y-%m-%dT%H:%M:%S.%fZ'
>>> var._properties
{'text': '43', 'updateTime': '2018-01-22T21:39:44.095040522Z'}

Observation: The %f format accepts microseconds, not fractions of arbitrary precision. That means six digits at most, but I see nine in the data.

[tswast]

Interesting. RuntimeConfig must have started recording nanoseconds somewhat recently.

@jonparrott It seems Python only supports microsecond precision, not nanoseconds.

I'm guessing we should modify the parsing to discard nanoseconds when present.

[tseaver]

@tswast FWIW, spanner has a non-public TimestampWithNanoseconds (derived from datetime.datetime). We could promote it if that would be useful.

[theacodes]

promote it into api_core? It might be useful because many APIs do expose nanosecond precision.

[tswast]

The from_rfc3339 function there looks very useful. I would bet that more GCP products will be doing nanoseconds.

[theacodes]

@tseaver do you mind filing a bug to promote that class to api_core? (& if you or @chemelnucfin
want to take on a PR to do that, that'd be awesome)

[tseaver]

@jonparrott see #4829.

[tseaver]

PR #4979 moved that class into API core, so we should be able to fix this issue using it.