Added R wrappers for released dataset ops

[terrytangyuan]

This PR is on top of #39. The wrappers were scaffolded using scripts created in #39 and then manually edited for cosmetic purposes as well as additional checks for arguments.

• Added R wrappers for released dataset ops, including kafka, kinesis, hadoop, and ignite. Note that only wrappers for the released dataset ops are included here - parquet and libsvm support are not included yet.
• Re-exported useful functions from tfdatasets package so tfio would play nicely with tfdatasets.
• Removed unnecessary dependencies and fixed R CMD check errors and warnings.

This is part of #6.

[terrytangyuan]

@yongtang Just rebased and resolved conflicts. Could you take a look when you get a chance?

[yongtang]

I don't have a lot of experience with R other than some brief knowledge many year ago. I tried locally but had the error:

> dataset <- sequence_file_dataset("/home/ubuntu/io/tensorflow_io/hadoop/python/kernel_tests/testdata/string.seq")
Error in sequence_file_dataset("/home/ubuntu/io/tensorflow_io/hadoop/python/kernel_tests/testdata/string.seq") : 
  attempt to apply non-function

I am wondering if there is a simple readme about the usage?

I think sequence_file_dataset is appropriate as an example, as it is pretty straightforward and only need a file name. (Kafka/Ignite requires additional setup).

[terrytangyuan]

Good call. I’ll add some test cases and instructions (maybe on top of your README PR). I am marking this PR as WIP for now and will ping when ready.

[yongtang]

Thanks @terrytangyuan. The repo have a simple sequence file:
https://github.com/tensorflow/io/tree/master/tensorflow_io/hadoop/python/kernel_tests/testdata

The sample sequence file consists of key-value pairs:

io/tensorflow_io/hadoop/python/kernel_tests/hadoop_test.py

Lines 37 to 41 in fc10234

	The file is generated with `org.apache.hadoop.io.Text` for key/value.
	There are 25 records in the file with the format of:
	key = XXX
	value = VALUEXXX
	where XXX is replaced as the line number (starts with 001).

Let me know if I can be of any help. I am interested in playing with R as well 😃

[terrytangyuan]

@yongtang Thanks. I fixed the error you encountered and added instruction to README.md as an alternative to the instruction you added in the other PR. Also added a very simple test case (we can add more strict ones later).

I haven't used tensorflow/tensorflow:custom-op image too often yet but the r-base image is used more often in the R community and solves some issues with certain R package dependencies that we don't have to deal with. We can investigate some time to make tensorflow/tensorflow:custom-op work too later on.

[yongtang]

@terrytangyuan I tried locally and it works. Thanks! We could add some additional test cases later.

[yongtang]

LGTM