Fix indexing issue for output monitoring

[Steven-Roberts]

Output monitoring functions would get called twice in a row with iter = 0 (unless dump_on_restart = false) and would never reach iteration NOUT which was used as a check for cleanup:

BOUT-dev/src/solver/solver.cxx

Line 931 in c7020b4

if (iter == NOUT || abort) {

This PR fixes solver iterations to be 1-indexed.

[dschwoerer]

Does this break the output? I think there are somewhere some tricky details, especially with smaller / slower output monitors, that make things a bit non-intuitive ...

[dschwoerer]

Why not change the check to NOUT - 1 instead of changing 19 files?

[ZedThree]

Does this break the output? I think there are somewhere some tricky details, especially with smaller / slower output monitors, that make things a bit non-intuitive ...

I'm also a bit wary about touching this stuff because I recall how fiddly the details are, but we do have pretty good tests on calling multiple monitors with different frequencies (SolverTest.AddMonitorCheckFrequencies for example), and that still seems to pass, so I'm in favour of merging this. It does also seem to make the logic easier to understand, even if do end up with some 1-indexing loops!

[dschwoerer]

The output, at least for non restarted simulations also seem to match, so might be good.

@Steven-Roberts could you please merge Steven-Roberts#1 to fix the formatting?
The CI is failing because it is from a fork, otherwise it would have found those issues.