Fix tag parsing bugs, and bring the project setup up to date

[timvink]

Bugs

Each fix has a regression test.

Two tags on one line crashed the build. The tag pattern matched its arguments greedily, so {{ read_csv('a.csv') }} and {{ read_csv('b.csv') }} became a single match that swallowed the text between the tags, and then failed to parse with SyntaxError: unmatched ')'.

An = inside an argument value raised a SyntaxError. parse_argkwarg() decided arg-versus-kwarg on any = in a segment, then split on every =. So read_csv('a=b.csv'), sep='=' and na_values=['a=1'] all failed. The same scanner tracked brackets and parentheses but not braces, so dtype={'a': 'str', 'b': 'int'} was split on its comma. Both now use one scanner that knows what is nested inside quotes, brackets, braces or parentheses.

convert_to_md_table() altered the DataFrame passed in. It escaped the columns in place before copying, so a mkdocs-macros-plugin user rendering the same DataFrame twice saw doubly escaped pipes in the second table.

Project setup

• Deploy the documentation on every push to master. It was deployed by hand, and the live site had fallen behind the readers added in 4.0.0 — read_parquet, read_orc, read_xml, read_html, read_stata, read_sas, read_spss and read_hdf were all missing.
• Upload coverage to Codecov again. The step tested env.USING_COVERAGE, which is set nowhere in the repository, so it never ran.
• Move dev dependencies to [dependency-groups], which uv now expects, and drop the unused codecov package.
• Add .codespellrc, which the codespell workflow already referred to.
• Add the enabled option to schema.json, so editors stop flagging valid configuration.
• Add CHANGELOG.md.
• Update the Makefile and CONTRIBUTING.md, which still described tests/test_requirements.txt, pyflakes and setup.py.
• Point the README's macros link at the hosted docs; the relative link 404s on GitHub and PyPI.

Version bumped to 4.0.1. Tests: 59 passing, 100% coverage.

🤖 Generated with Claude Code

https://claude.ai/code/session_017vTNbovqPphizF3v1UdhVW