Do not print file headers when restarting

[danielhollas]

Aside from being nicer, this will also enable to more easily test that we get the same results when restarting.

[codecov]

Codecov Report

Merging #106 (6851dec) into master (183ddcd) will increase coverage by 0.02%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #106      +/-   ##
==========================================
+ Coverage   87.10%   87.13%   +0.02%     
==========================================
  Files          41       41              
  Lines        5793     5805      +12     
==========================================
+ Hits         5046     5058      +12     
  Misses        747      747              

Impacted Files	Coverage Δ
src/init.F90	70.15% <ø> (ø)
src/tera_mpi_api.F90	99.40% <ø> (ø)
src/utils.F90	100.00% <ø> (ø)
src/vinit.F90	97.72% <ø> (ø)
src/files.F90	99.27% <100.00%> (+0.06%)	⬆️

[suchanj]

Well I did not mind having the headers there to remind me restart was needed. However with properly functioning code they should not be needed so feel free to go ahead.

[danielhollas]

@suchanj good point, that was also a reason why I didn't mind them. This commit makes it easier to process the files, since one can just skip the first line, and don't worry about anything extra. But my main motivation here was for tests.

Specifically, I wanted to be able to easily test whether restarted simulation gives the same result as the continuous simulation. Now I can do something like this:

mv tests/CMD/intput.in2 tests/MORSE/intput.in
sed -i 's/irest=1/irest=0/' tests/MORSE/input.in && rm tests//restart.xyz.*.ref
make test TEST=MORSE

and see whether the test passes. Previously the extra header lines would mess that up.
I already previously noticed a small issue with SH restart, will open an issue and fix next week.

Nevertheless, with the refactor done here we can easily go back if we want.