build: switch to pyproject.toml

[afuetterer]

Hi, since you merged my PR about the gha setup, I thought I might submit another one.

This modernizes the project setup using a pyproject.toml file.

What do you think?

Refs:

• https://packaging.python.org/en/latest/guides/writing-pyproject-toml/
• https://setuptools.pypa.io/en/latest/userguide/pyproject_config.html

Tasks

• Rebase
• Drop pyyaml from dependencies
• set requires-python to >= 3.10
• add 3.10-3.14 to trove classifiers
• add 3.10-3.14 ci tests
• check if license can be included without manifest file
• squash commits
• update coveralls and pytest-cov dependency (updated previously in requirements.txt)
• waiting for fix: replace urlretrieve in getRemoteFile #444 to be merged, so that CI can (hopefully) pass
• rebase
• squash commits
• add parallel coverage completion job
• remove mentions of easy install in readme
• test: replace url in test_remote_html test case #453
• ci: test multiple python versions with parallel coverage uploads #454
• rebase
• clean up ci.yml, remove install requirements.txt, remove pytest -ra --cov=tika
• squash commits

[chrismattmann]

If this PR looks good to you, I will squash all those tiny commits.

LGTM please squash.

[coveralls]

coverage: 55.646% (+2.2%) from 53.469%
when pulling 66aae33 on afuetterer:toml
into ce4c3ca on chrismattmann:master.

[afuetterer]

Failing test case due to network reasons.

[chrismattmann]

Failing test case due to network reasons.

Do we need the failing test to be fixed first? Looks like it's tied to a specific Python version @afuetterer

[afuetterer]

I don't think it depends on the Python version.

It is about reaching http://neverssl.com/index.html (port 80 without ssl), that results in an error. Maybe because I introduced verify_ssl in the new _urlretrieve function. But it worked during that PR, which confuses me.

[chrismattmann]

I don't think it depends on the Python version.

It is about reaching http://neverssl.com/index.html (port 80 without ssl), that results in an error. Maybe because I introduced verify_ssl in the new _urlretrieve function. But it worked during that PR, which confuses me.

OK interesting. We should probably make sure the build passes on that version...