Implement truth probability for boolean generator

[Kolyunya]

What was wrong

Currently, it's impossible to generate booleans that are biased towards True or False values.

How this fixes it

This patch adds an optional truth_probability argument to the pybool() method which allows to generate booleans that are biased towards True or False values.

Having the new argument is optional and defaults to 50%, this patch is fully backward compatible.

Usage example

Imagine we need to generate random booleans but we only want 10% of them to be equal to True. After this patch it would be possible using the following code:

boolean = faker.pybool(10)
boolean = faker.pybool(truth_probability=10)

Side notes

Our PHP friends already have this feature implemented which proves that it is actually useful for the end-users.

PHP implementation has chanceOfGettingTrue as the name of the argument, but I believe that truth_probability would fit better.

Seems like it's impossible to use parametrize in that test case because it extends unittest.TestCase, that's why I added a helper method to the test class.

[fcurella]

Thank you!

[Kolyunya]

@fcurella my pleasure!