Fix allow_none to work correctly on BaseInstance objects

[mdickinson]

Currently instances of BaseInstance don't clone themselves correctly when called with a new version of allow_none. This means that Time(allow_none=False) and Datetime(allow_none=False) don't work as expected:

Example (on the master branch):

>>> from traits.api import *
>>> Slice = BaseInstance(slice)
>>> SliceNotNone = Slice(allow_none=False)
>>> Slice.validate(None, "dummy", None)  # accepts None, as expected
>>> SliceNotNone.validate(None, "dummy", None)  # accepts None, but shouldn't

This PR fixes the issue by extending the base class clone method to add special handling of allow_none. With the PR, the above example becomes:

>>> from traits.api import *
>>> Slice = BaseInstance(slice)
>>> SliceNotNone = Slice(allow_none=False)
>>> Slice.validate(None, "dummy", None)  # accepts None, as expected
>>> SliceNotNone.validate(None, "dummy", None)  # rejects None, as expected  
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/Users/mdickinson/Enthought/ETS/traits/traits/trait_types.py", line 3361, in validate
    self.validate_failed(object, name, value)
  File "/Users/mdickinson/Enthought/ETS/traits/traits/trait_types.py", line 3211, in validate_failed
    self.error(object, name, value)
  File "/Users/mdickinson/Enthought/ETS/traits/traits/base_trait_handler.py", line 75, in error
    object, name, self.full_info(object, name, value), value
traits.trait_errors.TraitError: The 'dummy' trait must be a slice, but a value of None <class 'NoneType'> was specified.

We also take the opportunity to add some basic tests for the Datetime and Time trait types.

Fixes #495

Checklist

• Tests
• [ ] Update API reference (docs/source/traits_api_reference) N/A
• [ ] Update User manual (docs/source/traits_user_manual) N/A
• [ ] Update type annotation hints in traits-stubs N/A

[mdickinson]

Note: while this is a bugfix, it's a long-standing bug that hasn't caused major issues, so it doesn't seem worth backporting to any bugfix release on the 6.2.x branch.

[mdickinson]

Drive-by documentation fix; no associated change in behaviour.

[rahulporuri]

LGTM with one question - I'm not sure why this fix only works on BaseInstance objects and not on BaseInstance subclasses? Won't the subclasses inherit the behavior via the inherited clone method?

[mdickinson]

I'm not sure why this fix only works on BaseInstance objects and not on BaseInstance subclasses?

It works on both. I'll fix the title to be less misleading. (The point of the title is that the usual pattern is some_trait_class(allow_none=True), where allow_none is a parameter to the initializer of the TraitType subclass, but the bug we're fixing is for the case of trait_instance(allow_none=True), where trait_instance is already an instance of BaseInstance.)

[mdickinson]

To be clear: before this PR

• If SomeTraitType is a subclass of BaseInstance, then foo = SomeTraitType(allow_none=False) already works as expected
• If SomeThing is an instance of BaseInstance (as e.g., Datetime and Time are), then foo = SomeThing(allow_none=False) does not work as intended - None is still allowed

This PR fixes the second part above.