Added full Markov-model pipeline

[PhilippSchmelter]

Closes #10

This pull request introduces a comprehensive implementation of a Markov-model pipeline for analyzing time-series data, including preprocessing, transition matrix generation, and testing. The most significant changes include adding the Markov-model pipeline to the project, implementing bucket-based time mapping, and updating the loader and tests to integrate the new functionality.

Markov-model pipeline implementation:

• Added the full Markov-model pipeline, including transition count and probability matrix generation. (CHANGELOG.md, CHANGELOG.mdR11)
• Created src/markov/transition_counts.py and src/markov/transitions.py for generating raw transition counts and Laplace-smoothed transition probabilities, respectively. [1] [2]

Time-bucket mapping:

• Implemented a bucket-based time mapping system in src/markov/buckets.py, which maps timestamps to unique bucket IDs based on month, weekend flag, and quarter-hour intervals. (NUM_BUCKETS constant and bucket_id function)
• Integrated bucket assignment into the preprocessing pipeline by modifying the load_timeseries function to add a "bucket" column. [1] [2]

Refactoring and modularization:

• Refactored and modularized the Markov-related functionality by creating a dedicated src/markov package with an __init__.py file for better organization and reusability.
• Added _core.py to encapsulate shared Markov-model logic, such as the _transition_counts function.

Testing and validation:

• Updated tests/test_loader.py to validate the integration of the Markov-model pipeline, ensuring the loader correctly adds "state" and "bucket" columns and computes expected values.
• Enhanced the test coverage to include validation of all new columns introduced by the preprocessing pipeline, such as "scaled," "state," and "bucket."

Miscellaneous:

• Removed unused code and cleaned up src/main.py and src/markov/model.py. [1] [2]