fix: fix as_dataframe

[busunkim96]

The dataframe helper methods were being skipped because pandas was not installed in the unit test session. This re-enables those tests and makes the minimal number of fixes to get the unit tests passing again.

TODOs:

• Are datetimes being handled correctly?
• Is there a better way to do WhichOneOf with proto-plus? Document how to use module_pb2.Message.WhichOneof proto-plus-python#137

Fixes #90 🦕

Simple example:

import datetime

from google.cloud import monitoring_v3
from google.cloud.monitoring_v3.query import Query
import pandas

# TODO: Change these to the desired project and metric_type
project_name = "my-project"
metric_type = "storage.googleapis.com/storage/object_count"

client = monitoring_v3.MetricServiceClient()

query = Query(client, project_name, metric_type=metric_type, end_time=None, days=7,)

df = query.as_dataframe()
print(df[:3])

[busunkim96]

Ah what this seems to do is convert the time to just nanoseconds... https://proto-plus-python.readthedocs.io/en/latest/reference/datetime_helpers.html#proto.datetime_helpers.DatetimeWithNanoseconds

[busunkim96]

pandas seems to have its own timestamp type (https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.Timestamp.html), so I'm not sure how these tests used to work 🤔

[parthea]

LGTM. Just one minor observation.