Add new functionality to calculate timestamps for dataframe splits based on user-defined percentages

[WladRamos]

This PR introduces a new function, calculate_timestamps, which allows splitting a dataframe into train, validation, and test sets based on percentage inputs. Additionally, the function supports optional preservation of continuous periods, determined by a time tolerance parameter.

Key Features:

• Percentage-based splits: The function takes percentages for train, validation, and test sets and returns the corresponding timestamps for splitting the data.
• Period preservation: When the preserve_periods option is enabled, the function ensures that the splits occur at the end of continuous periods, preventing data from being split in the middle of a period.
• Time tolerance: The function includes a time_tolerance parameter, which defines the minimum gap between observations to be considered as the start of a new period.

Unit Tests:

• Added comprehensive unit tests to cover:
  • Period preservation behavior.
  • Validation of input percentages (ensuring they sum to 1).
  • Handling of dataframes without a DatetimeIndex.
  • Small time_tolerance cases to ensure gaps are correctly respected.