Make unneeded count operation optional in clean-duplicate-history.

[bigtimecriminal]

Problem Statement
Performance of the clean-duplicate-history function could be optimized.

Describe the solution you'd like
I want to suggest a simple improvement. There's nested count operation the result of which is only using in a log statement at verbosity level 2: https://github.com/jazzband/django-simple-history/blob/master/simple_history/management/commands/clean_duplicate_history.py#L69

I'd like to run count only if we actually log the result. Otherwise we should use exists.

Describe alternatives you've considered
None. I think this is a pretty simple change