#34 - Experiment with analyzer integration tests

[Kamforka]

As per #34

The difference to the existing integration test suite is that this approach is relying on the actual input.json/output.json setup, instead of just piggy backing on the stdin/stdout approach which is not what Cortex is relying on in real life.

These 6 sample tests alone cover a large portion of the lib:

$ pytest tests/test_integration_analyzers.py --cov
====================================== test session starts ======================================
platform linux -- Python 3.14.4, pytest-9.0.3, pluggy-1.6.0
rootdir: /home/ubuntu/projects/cortexutils
configfile: pyproject.toml
plugins: cov-7.1.0
collected 6 items                                                                               

tests/test_integration_analyzers.py ......                                                [100%]

======================================== tests coverage =========================================
________________________ coverage: platform linux, python 3.14.4-final-0 ________________________

Name                       Stmts   Miss  Cover
----------------------------------------------
cortexutils/__init__.py        3      2    33%
cortexutils/analyzer.py       70     27    61%
cortexutils/extractor.py      68     10    85%
cortexutils/worker.py        113     36    68%
----------------------------------------------
TOTAL                        254     75    70%
======================================= 6 passed in 0.09s =======================================