Upgrade project configuration and QA tools

[dvp2015]

The numjuggler package configuration needs updating. There are problems with creating a debugging environment with modern libraries.

The project doesn't use any linting, formatting, typing and other QA tools, without which contribution is if not possible, but can be too risky or produce a lot of changes to review.

To my opinion, the project is a great work, it took a lot of efforts in the past and it is still valuable. But the project needs some update.

In this change I have:

• dropped support of Python2 in favor of using newest libraries
• allowed Python 3.11..14 (as the recent numpy does)
• tested on Python 3.14
• moved configuration and setup from setup files to pyproject.toml
• changed layout to src-layout to comply with recent recomendations
• reimplemented GitHub Acion to test an all the Python versions, now GHA is based on uv and pyproject.toml
• fixed tests and added one new for the bug I've found recently (XFAIL for now, will be fixed later)
• add dependencies for code QA: ruff, ty, pylint
• added justfile for some routine development task.

The travis tests now are implemented as simple pytest tests.

I used uv with this project. It's really efficient and quite simple in use. In 99% of cases one only need to know only the following commands:

uv venv
uv sync 
uv pip 
uv add/remove
uv run

If you use direnv there's also appropriate .envrc in the project root to automatically activate the venv.

[dvp2015]

The old travis tests are reimplemented in pytest. The tests found some bugs on Windows which require to test and revise the parser module. The test coverage increased from 20% to 30%. Should be at least 80% for (never safe) refactoring. One more advantage: now the tests can be debugged.