Clean-up main model: migrate reset calculation info to job dispatch

[mgovers]

The location where the calculation info was reset is quite opaque. To more explicitly show the calculation

NOTE: this does not solve the actual issue that main model owns its own info, nor the fact that the location where the reset is done is somewhere halfway through the calculation. It merely makes it more visible what happens under the hood.

• Run Benchmark and make sure that there are no regressions like the one introduced in Refactor calculation-info: timer using codes #1087

Relates to #1082 (reinstigated the original logic to allow benchmarking against original main)
Relates to #1093 (this PR is an enabler for follow-up improvements similar to #1093, but it is not a blocker for it)

[mgovers]

still something weird going on with the benchmark. do-not-merge label added

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
100.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[mgovers]

still something weird going on with the benchmark. do-not-merge label added

Found the bug on main that caused the new results to (incorrectly) appear much slower compared to the original one:

However, after digging into this, it turns out that the tap position optimizer contains a bug in the following line: https://github.com/PowerGridModel/power-grid-model/blob/feature/migrate-reset-calculation-info/power_grid_model_c/power_grid_model/include/power_grid_model/optimizer/tap_position_optimizer.hpp#L1033

On current main, the calculation info was cleaned up before every tap position optimizer subcalculation, and, as a result, the tap position optimizer was not fully benchmarked, because the intermediate timings were discarded - not great. Moving cleaning responsibility to the JobDispatch meant that we now DO benchmark the tap position optimizer, effectively fixing this bug.

NOTE: this also relates to #1082 .

TBD: we have 3 paths going forward:

• merge Clean-up main model: migrate reset calculation info to job dispatch #1094 as-is, including the fix for the calculation info. that means that the benchmarks on versions after this PR will not be compatible with versions prior to this PR.
• add an info.clear() line in that location in Clean-up main model: migrate reset calculation info to job dispatch #1094. that would actually just shift the problem from main model impl to the optimizer. i don't particularly like this solution.
• same as 2. but immediately after that PR, do a fix that removes the info.clear(). that way, there is a single PR that contains a single line bugfix, but we still have the downsides of 1.

[mgovers]

Decision: merge this PR as-is and recalibrate our benchmark.

[mgovers]

still something weird going on with the benchmark. do-not-merge label added

do-not-merge label removed cfr. above