TaskRunner E2E: Parameterised input and coverity fix

[noopurintel]

Parameterised the input for job, model and python version in TaskRunner E2E for better manual triggers.

Also, added a check on the fetched value of GITHUB_STEP_SUMMARY before using it - as part of Coverity fix.

Successful E2E job runs:

1. All tests + keras_cnn_mnist + python 3.12 -> https://github.com/noopurintel/openfl/actions/runs/12922201062
2. Test without TLS + torch_cnn_mnist+ python 3.11 -> https://github.com/noopurintel/openfl/actions/runs/12922184838
3. All tests + models + python versions (just like existing flow) -> https://github.com/noopurintel/openfl/actions/runs/12922444478