Inconsistent syntax expectations in data files

[gilles-peskine-arm]

We have two pieces of code that parse .data files containing unit tests:

• framework/scripts/mbedtls_framework/test_suite_preprocessor.py, moved from framework/scripts/generate_test_code.py in Generate MLDSA test cases for the driver and dispatch layers #282, used to generate .datax files which are then used to run the unit tests.
• framework/scripts/mbedtls_framework/collect_test_cases.py, used in particular to list the test cases in outcome analysis, so it's what we use to check that all test cases are executed.

These have independent parsers, and they make different assumptions when we deviate from our habits. In particular:

Description 1
function1:argument1
Description 2
function2:argument2

without a blank line is treated as two test cases when generating .datax file, but “Description 2” is ignored for coverage analysis, so we may miss some test cases.