feat: add pyproject.toml for uv/setuptools wheel builds

Added heuristics-based transfer of linguistic knowledge based on word alignment. Before that, word alignment also improved using linguistic knowledge based on morphologicand, POS and dependecy parse from the source language to reduce NULL alignments for example, so as to correct obvious omissions from alignments. Updated requirements files, YAML files,docs, CLI and usage for this extension. Effectively merges the SyntheticWrdAlignedUDTB project into this project, so that it does not just word alignment and provides data structure API (with SSF and CoNLL-U support), but also aligns and creates treebanks (not necessarily synthetic).

Extended docs to include details about the new options for scoring etc..

Added alignment scoring using some common ways. Added these options to logging, CLI and usage.

Integrated the ability to use any BERT model such as xlm-roberta-large or indic-bert-v2 etc. Improved the CLI and usage help in the main run scription.

CLI improved with better usage and help options.

Version 3.0 working. Added word alignment with awesome-align and fast_align, converter from their output to GIZA++ .A3.final files. Some docs and tests added.

Version 2.0 working.

Version 1.0: Fully functional API for aligned linguistic units, as well as awesome-align based word alignment. Includes two notebooks for each of them.

Initial commit