Improve integration tests: compare SNOWGLOBES output files for pull request

[Sheshuk]

Working on #126 #127 I need to validate the output of our snowglobes interface, comparing it with the results from main branch. An idea (inspired by the existing integration test) is to make an automated pipeline for this using github actions.

• Run the full chain for specific model/models
• For the main branch - store the resulting archive (using upload-artifact action).
• For other branches - retrieve the archive (using download-artifact action) and compare the contents of the file to output.

In the same way we can also run the timing benchmarks and store the results for comparison, like described here

[Sheshuk]

Hmm, looks like you can't reuse an artefact from another run. In that case I'll just include a reference file with a table to use.

[Sheshuk]

Integration tests implemented in #132 Using the table from SNEWSv2 paper for the event numbers cross-check