CURA-13249 Merge header with first GCode part

[Erwan MATHIEU (wawanbreton)]

Instead of sending the GCode header as a separate part, we now send it with the very first part of the actual GCode commands, so that it matches the previous behavior and should allow the post-processing scripts to work without having to fix them all.
Since I am not very familiar with the post-processing scripts, I assume this is the correct fix but the result looks a bit weird to me, so please double-check.

⚠️ This PR merges to CURA-13226_slice-speed-benchmark-tooling because it also changes the behavior of the initial GCode sending

CURA-13249

[github-actions]

Test Results

31 tests  ±0   31 ✅ ±0   5s ⏱️ ±0s
 1 suites ±0    0 💤 ±0 
 1 files   ±0    0 ❌ ±0 

Results for commit 666e6fa. ± Comparison against base commit 5a53fc2.

[Casper Lamboo (casperlamboo)]

Don't fully understand why this is needed, but code looks good

[Erwan MATHIEU (wawanbreton)]

Don't fully understand why this is needed, but code looks good

The front-end saves the GCode parts in a list, and the post-processing plugins expect this list to be in a certain order, with the first elements containing specific parts, namely the header, init gcode and start gcode. With the new structure we had an intermediate part inserted, so scripts were confused.

Other option would be to change the post-processing scripts, but that would require more effort, and possible be annoying because the start GCode is optional.