[rozyczko]

Data consistency and type improvements:

• Changed the representation of boolean fields in layer and assembly dictionaries from string values (e.g., 'True'/'False') to actual Python booleans, improving type safety and clarity (_from_layers_collection_to_list_of_dicts, _from_assemblies_collection_to_list_of_dicts, and related usages).
• Updated the layers property and related functions to use more precise type annotations, reflecting the inclusion of boolean values.

Minimizer and Bayesian logic refactoring:

• Refactored minimizer selection logic to simplify the list of available minimizers, clarify the handling of the Bayesian option, and ensure correct index mapping between GUI and backend.
• Exposed VALID_INITIALIZERS for Bayesian sampling as a public constant and updated all references to use it, ensuring consistency between frontend options and backend validation.

Parameter filtering and display improvements:

• Simplified experiment/model parameter filtering by using the explicit parameter['experiment'] field instead of helper methods, reducing indirection and potential errors.
• Improved logic for generating parameter display names, especially for layer and magnetism parameters, by refining path handling and helper functions.

Plot rendering and result handling:

• Updated Bayesian plot property methods to clarify that plots are rendered after sampling finishes, not on property read, and removed unnecessary on-demand rendering logic.

Other codebase cleanups:

• Improved the add_new method in layers logic to return whether a new material was added, increasing transparency for the caller.
• Added missing imports and clarified type annotations for better code clarity and reliability.