fix: benchmark names with multiple lines

[gamma0987]

Here's the fix for the problem on discord https://discord.com/channels/1074890101362004088/1412794833072033903/1413138826645209198 and as discussed in #619. Because there's no github issue: Here's a quick summary of the problem and the solution:

In the output of the slang benchmark there are benchmarks which have a multiline benchmark name. These benchmarks are not parsed correctly and are skipped by the current parser which causes the discrepancy between the expected amount of 12 benchmarks and the actual amount of 7 in this benchmark run https://github.com/NomicFoundation/slang/actions/runs/17408413743/job/49418989630. This issue arises in gungraun due to the use of syn's to_string function, which occasionally splits strings across multiple lines. While this behavior may not be considered a bug within gungraun, it deviates from the original intention of keeping both the benchmark name and description on a single line.

This pr contains a fix for the parser to accept benchmark names on multiple lines. Multiline benchmark names and descriptions will be removed in future versions of gungraun but this fix should work for all versions of gungraun/iai-callgrind.

[gamma0987]

Lot's of errors in the ci. Can I do something about it?

[epompeii]

@gamma0987 thank you for PR!
This all looks good to me. The CI failures are expected. I need to update the workflows to be more fork PR friendly.