support for an tf.io.gfile.GFile like API

[kpe]

It would be nice to support a tf.io.gfile.GFile like API because it can be used as a drop-in replacement for the normal pythonic open() API to transparently handle both local and Google Cloud Storage files (i.e. by using the gs:// schema).

Is your feature request related to a problem? Please describe.
Well it is much more convinient to use a familiar API like:

filepath = "gs://my-bucket/gcs-file.txt" if True else "/tmp/local-file.txt"
with gcs.GFile(file_path) as f:
  print(f.read())

then considering anything like:

#
# how exactly should I read the content of gs://my-buckets/gcs-file.txt 
# - where's the docu?
# - should I read it all?
# - I just need to read a text file (from GCS)
#
from google.cloud import storage
from google.cloud.storage.blob import Blob
client = storage.Client()
blob = Blob.from_string("gs://my-buckets/gcs-file.txt")
blob.download_as_*** # ... agghhh.h.. I'll just give up

Describe the solution you'd like
The way acess to GCS buckets is handled in TF (i.e. tf.io.gfile.GFile) is just perfect, however if I only need GCS file access, I may not want to install all of tensorflow.

Describe alternatives you've considered
As an alternative one could try splitting tf.io.gfile.GFile into a standalone pip package.

I'm not sure if this is the right place for a feature request like that, but it would be nice if there is a some pyton ideomatic open() like API. And if it were there, I guess, it would be used 90% of the time.

[frankyn]

Thanks @kpe, this is a duplicate of: #29

@andrewsg is working on a solution.