Add options for UNKNOWN state's distribution in BoundedQNT

[iturbideM]

Description

Currently, when an unknown state is created in BoundedQNT, its distribution is computed as the mean distribution of all paths leading to it. This behavior is controlled by the compute_mean_distribution_for_partial_hypothesis flag. To improve flexibility, we should refactor this implementation to support alternative methods for calculating the distribution of the UNKNOWN state.

Proposed Changes

• Refactor BoundedQNT to replace the flag with a more flexible mechanism for defining the distribution of unknown states.
• Document the new approach and its usage.
• Implement alternative methods for calculating the distribution of UNKNOWN states.