Use a defined sourceTime when using parsedatetime

[nicoe]

When using the parsedatetime functionalities freezegun is not used to
compute the source time. If the tests are run too close to midnight it
might result in the computation of 'tomorrow' being another day then
now() + 24h (because time.localtime() will be the next day) … thus
failing test_default_due2.

Setting sourceTime when calling parsedatime will ensure the same base is
used for all tests.

[nicoe]

AFAICT this should fix #320

[WhyNotHugo]

I'd rather decorate affected tests, and keep non-test code as is in this case.

Do you know exactly which tests are affected by this?

[nicoe]

* Hugo Osvaldo Barrera [2019-08-26 16:17:23]:

I'd rather decorate affected tests, and keep non-test code as is in this case. Do you know exactly which tests are affected by this?

test_default_due2 and I just noticed that it fails also at this time and hour when the timezone is CEST but on the second test: assert todos['ccc'].due == todos['bbb'].due - datetime.timedelta(hours=23) (it does not fail with "my" patch though)