Adding primitive Bigtable garbage collection rule classes.

[dhermes]

These are used as settings when creating new column
families within a table.

Also slipping in 3 small Bigtable docstring updates.

[dhermes]

I am assuming (and am re-implementing under the assumption) that you are more in favor of 13+ subclasses than a class with 13+ arguments (for the filter case). (Here it's simpler since there are only two options.)

[tseaver]

I am assuming (and am re-implementing under the assumption) that you are more in favor of 13+ subclasses than a class with 13+ arguments (for the filter case). (Here it's simpler since there are only two options.)

Exactly, because their implementations will be super-simple / obvious. For the filters, It is also a lot clearer to map the separate classes onto a "syntax tree" when we get around to intersection / union logic.

[dhermes]

@tseaver PTAL.

I used _total_seconds from:
https://github.com/GoogleCloudPlatform/gcloud-python/blob/ea4ebb7b87b9ff27ac9cae9cd7c0e5d3ad630dea/gcloud/bigquery/_helpers.py

[dhermes]

Looks like https://pypi.python.org/pypi/protobuf/3.0.0b1.post1 (released today) caused Travis to fail:

...
Collecting protobuf>=3.0.0a3
  Downloading protobuf-3.0.0b1.post1.tar.gz (200kB)

[dhermes]

@tseaver I filed protocolbuffers/protobuf#1056 and pinned to 3.0.0a3 for now. If you like I can pin to it in another PR.

[tseaver]

I used _total_seconds from:
https://github.com/GoogleCloudPlatform/gcloud-python/blob/ea4ebb7b87b9ff27ac9cae9cd7c0e5d3ad630dea/gcloud/bigquery/_helpers.py

Shouldn't we remove that version, and replace it with the new one in gcloud._helpers?

Other than that, LGTM.

[dhermes]

@tseaver It's not used anymore, I just remembered we had it before we put all time related stuff in gcloud._helpers.