Clean up timedelta/relativedelta usage

[arkid15r]

• Standardize relativedelta days offset usage
• Replace timedelta with relativedelta

[coveralls]

Coverage: 100.0%. Remained the same when pulling 56ac25a on arkid15r:relative-delta-clean-up into 84af05e on dr-prodigy:beta.

[dr-prodigy]

Hi @arkid15r , I like the idea behind this PR a lot! although I have an important note to add: given this refactoring is very large, I would prefer to have it split in 2, ie:
1/ business logic code refactoring, no changes on tests
2/ no changes on BL, tests refactoring
so to guarantee no regressions without requiring extensive visual reviews.
Do you think it's still possible?
Thx a lot!

[arkid15r]

That's a great point, @dr-prodigy!

I implemented and tested these changes in the 2 steps you've just described. However, the PR doesn't reflect that.
It makes perfect sense to split it into 2 as you suggested. I've extracted tests into https://github.com/dr-prodigy/python-holidays/pull/894

Thanks for the review!