Sharrow compatibility for v15.3.1

[jpn--]

This pull request introduces new scripts and configuration files to support running the SANDAG ABM3 example model at different scales and with different settings, including sharrow and multiprocessing options. It also updates dependency requirements and makes minor corrections to configuration and sample files to improve consistency and avoid variable conflicts. The most important changes are grouped below.

New scripts and workflow enhancements

• Added several new scripts (run-small-sharrow.py, run-large-sharrow.py, run-large-legacy.py, run-large-sharrow-mp.sh, run-large-legacy-mp.sh) to enable running the model with different configurations (small/test, large/full-scale, sharrow enabled/disabled, multiprocessing) and improved reproducibility via embedded metadata for dependency management. [1] [2] [3] [4] [5]
• Added scripts/fulldata.py to handle downloading and verifying the full-scale dataset, including checksum validation and extraction, ensuring large datasets are managed efficiently and securely.
• Added a scripts/README.md explaining the purpose and usage of each script for easier onboarding and clarity.

Configuration updates for sharrow and multiprocessing

• Introduced a new configuration file settings_mp_sharrow.yaml to support running the model with sharrow and multiprocessing, including settings for process count, fail-fast behavior, and model steps.

Dependency and compatibility improvements

• Updated the required activitysim version in pyproject.toml from 1.4 to >=1.5 to ensure compatibility with newer features and bug fixes.

Data and configuration corrections

• Corrected variable references in trip_destination_sample.csv and trip_destination_annotate_trips_preprocessor.csv to avoid conflicts and ensure proper evaluation of expressions. [1] [2]
• Renamed the first column in data/maz_maz_bike.csv from Unnamed: 0 to Nothing for clarity.

Testing adjustments

• Relaxed the relative tolerance (rtol) in the model test checkpoint comparison to reduce false negatives in cross-platform testing.

[i-am-sijia]

@jpn-- Looks good, thanks! If I merge this, PR ActivitySim#39 becomes a duplicate of PR ActivitySim#37. I have no particular preference on which one we want to keep.