only call truncate if truncation is modelled

[sbfnk]

This simplifies the call to truncate in estimate_infections, only calling it when truncation is actually required. This seems more intuitive as the call to truncate was previously required even if no truncation was applied, but only to create an obs_report variable not containing the forecasts. This now follows the same logic as e.g. only applying scaling if obs_scale is set to 1, such that the decision of whether truncation is applied is now happening outside the truncate function, which now always truncates.

As a (positive, in my opinion) side effect this does away with having to pass vectors to the truncate function requiring declarations of vectors of size 1 in estimate_truncation.

[seabbs]

Looks good to me. The downside is more if/else shenanigans in the main model code potentially making it harder to track what is happening. Note should really have a news note or other indication of code changes.