Return non-zero exit status when validation fails

[chupakobra6]

Closes #22.

The CLI currently reports failed validation while returning exit status 0, so CI systems treat rejected datasets as successful checks.

Return an explicit status from main() and propagate it through sys.exit() at both CLI entry paths:

• 0 when every validated URI passes;
• 1 when any validated URI fails;
• 2 for argument and unrecognized-command errors.

The status is computed across every executed ValidateCommand. Per-URI and summary reports are still generated before returning, and the existing final stdout line listing failed URIs is unchanged. Document the exit-status contract in the README.

Add regression coverage for successful validation, mixed multi-URI validation, report generation on failure, failed-URI output, argparse errors, and the real python -m imas_validator entry point with an invalid netCDF dataset.

Validation on the rebased current head:

• Full Python 3.13 suite: 189 passed, 2 existing skips.
• Focused CLI scenarios: 5 passed.
• Black, Flake8, MyPy, and isort pass.

[olivhoenen]

Finding from Claude:

The check assert "FAILED validation." in completed.stdout depends on how long the temp path is. The "URI has FAILED validation." line is printed by rich's RichHandler. When stdout isn't a terminal, rich wraps lines at 80 columns by default, and it can break between "FAILED" and "validation.". That happens or not depending on how long the pytest temp path is.

It can therefore fail on CI, under -n=auto (whose worker subfolders make the path longer), or for users with longer usernames. Possible fixes:

• drop this check;
• or match on a normalised string, for example " ".join(stdout.split()).

[olivhoenen]

Thanks!