- Adds unit tests
- Also adds credits to original code author

Adds a test that runs the executor module's Beam pipeline using
a DirectRunner and exports prediction data to an actual BigQuery table.

Changes:
- Refactors the component integrate test and adds a test to run the
  component on Vertex AI Pipelines.

- Adds a Dockerfile to package the component code into
  a Docker image based on OSS TFX image. The image can then be used
  as the base image when running a pipeline in Vertex AI.

- Updates the `bigquery_export` output of the predictions-to-bigquery
  to store the generated BigQuery table name. This aids with checking
  the output of the component during testing, but also allows any
  downstream component receive this component's output.

Adds a test that integrates the transform component into the pipeline.
Test is implemented for local runner only.

Adds a container component stub to represent the TFX Transform component
for integration testing on Vertex AI.

Replaces create_tempfile and create_tempdir calls from abseil's
absltest.TestCase and parameterized.TestCase with equivalent
methods from the tempfile package.

The reason is that the abseil methods require parsing of the
FLAGs variable, which may not be executed if absltest.main() is
not invoked. This can happen when test filtering is performed, e.g.
```
python -m unittest path.to.test
```

Mentions the predictions-to-bigquery component in top-level readme.

- Fix issues in pred2bq readme
- Reverted version change in setup.py
- Add abls-py test prerequisite in setup.py