Support strict boundaries even when transform strategy is used

[amontanez24]

Problem Description

As a user, it would be useful if I could define range or inequality constraints without having to depend on reject sampling.

Expected behavior

Currently, the strict parameter in the range and inequality constraints can only be enforced using reject sampling. It is not guaranteed to work when set to True if the handling_strategy is set to transform.

• Improve both constraints to support the strict parameter being set to True when using the transform strategy.

Additional context

• Currently the strict parameter is only used to set the operator parameter. This is only used in the is_valid method which is only used to validate and during reject_sampling:
  https://github.com/sdv-dev/SDV/blob/master/sdv/constraints/tabular.py#L368-L371
  https://github.com/sdv-dev/SDV/blob/master/sdv/constraints/tabular.py#L454
  https://github.com/sdv-dev/SDV/blob/master/sdv/constraints/tabular.py#L741
  https://github.com/sdv-dev/SDV/blob/master/sdv/constraints/tabular.py#L845-L849
• We may want to use it to make sure that any values that are equal to the boundary are shifted up to the next acceptable value if strict is True
  • This is tricky because we will need to know if the next acceptable value is an int or float, and how many places it can be rounded to.