Fix docstring example syntax errors and test Python 3.9

[sadielbartholomew]

Closes #132.

Now the linting pre-commit checks are passing after #194, I want to check if Python 3.9 passes the CI to confirm the resolution of #132. It passes for me locally as far as I can tell around the seg faults which are still occurring and under investigation (see #180), though the seg faults may also appear on the Actions jobs since we have seen them emerge there before.

Since the Actions jobs for the workflow to run the test suite are only set to be triggered by PR, I am putting in a trivial PR to trigger and inspect the jobs. These trivial commits fix all of the syntax errors raised by doctest on the codebase (I have set up a CLI option -d so we can check these via py run_tests.py -d towards #4, though until they are all passing I won't add them to the test suite).

[sadielbartholomew]

Results are in... both the standard codebase-based (with 3.9) and the source dist (with 3.7) test jobs have failed (else been cancelled) due to one test method error emerging due to some spurious partitioning-based issue:

======================================================================
ERROR: test_Field_indices (test_Field.FieldTest)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/runner/work/cf-python/cf-python/main/cf/test/test_Field.py", line 1490, in test_Field_indices
    indices = f.indices(grid_longitude=cf.wi(50, 130))
  File "/home/runner/work/cf-python/cf-python/main/cf/field.py", line 13194, in indices
    b = self.flip(axis).anchor(axis, anchor1, dry_run=True)[
  File "/home/runner/work/cf-python/cf-python/main/cf/decorators.py", line 64, in precede_with_kwarg_deprecation_check
    operation_method_result = operation_method(self, *args, **kwargs)
  File "/home/runner/work/cf-python/cf-python/cfdm/cfdm/decorators.py", line 49, in inplace_wrapper
    processed_copy = operation_method(self, *args, **kwargs)
  File "/home/runner/work/cf-python/cf-python/main/cf/field.py", line 15527, in anchor
    if dim.increasing:
  File "/home/runner/work/cf-python/cf-python/main/cf/dimensioncoordinate.py", line 259, in increasing
    return self.direction()
  File "/home/runner/work/cf-python/cf-python/main/cf/dimensioncoordinate.py", line 359, in direction
    _direction = self._infer_direction()
  File "/home/runner/work/cf-python/cf-python/main/cf/dimensioncoordinate.py", line 129, in _infer_direction
    data = data[0:2].array
  File "/home/runner/work/cf-python/cf-python/main/cf/data/data.py", line 7921, in array
    p_array = partition.array
  File "/home/runner/work/cf-python/cf-python/main/cf/data/partition.py", line 1153, in array
    p_data = p_data.reshape(p_shape)
ValueError: cannot reshape array of size 9 into shape (1,)

----------------------------------------------------------------------
Ran 304 tests in 1246.062s

FAILED (errors=1)

@davidhassell should we be concerned about that error now, or assume it is some unknown facet of the partitioning that has been there for a long time as I believe we agreed last time?

The great news is no seg faults are happening (at least for the jobs that ran to completion and for the tests that ran before the cancellations otherwise)! Which is great as even with faulthandler it's difficult to diagnose and debug those.

Since the 3.9 test is passing beside the above test method in a way that is consistent with the jobs with the other minor Python versions, I am satisfied that 3.9 is running fine on the CI as well as locally. Hence:

• set this PR to close Verify support for Python 3.9 (once it is more established) #132 (via the opening comment);
• test completed so this PR is now good to merge.