Fix tests for get_standard_metadata

[luandy64]

Description of change

• Add docs
• use unittest methods to do asserts
• adds a missing test
• update expected values

This duplicates changes in #132, but I added the doc string to explain the rational behind the test, and the self.subTest() context manager to make the process of examining a failing test more sane.

Example test run:

$ python -m unittest tests/test_metadata.py
..
======================================================================
FAIL: test_standard_metadata (tests.test_metadata.TestStandardMetadata) (test_number=10)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/opt/code/singer-python/tests/test_metadata.py", line 321, in test_standard_metadata
    self.assertDictEqual(expected_value, actual_value)
AssertionError: {(): {'inclusion': 'foo', 'forced-replication-method': 'INCREMENTAL'}} != {(): {'forced-replication-method': 'INCREMENTAL'}}
- {(): {'forced-replication-method': 'INCREMENTAL', 'inclusion': 'foo'}}
?                                                 --------------------

+ {(): {'forced-replication-method': 'INCREMENTAL'}}

----------------------------------------------------------------------
Ran 3 tests in 0.001s

FAILED (failures=1)

Because it's divided into sub tests now, we get an error message like this for each failing sub test. And the (test_number=10) in the first line tells you which input is failing. Just grep for test_number=10 in the test file.

Perhaps a better approach would be to move contants into global constants and break the 1 massive test into the 16 individual tests, but @Jude188 did all of the adjustments already, so I left it that way.

Manual QA steps

• Ran the tests locally and they pass

Risks

• Low as the function didn't change
  • The tests had bad expected values and they were never caught because of the way the asserts were written

Rollback steps

• revert this branch