Simplify A+ cohort handling: positive contiguous IDs instead of negative-ID namespace

[pletzer]

No description provided.

[pletzer]

A+ cohorts previously lived in a separate negative-ID namespace with hand-rolled arithmetic (-timeStep-1, aPlusId+1, etc.), forcing every consumer to branch on task_id < 0 / std::abs(task_id). This PR makes A+ cohorts ordinary one-step cohorts appended right after the last normal ID (firstAPlusId = numAgeGroups + numTimeSteps - 1), exposed via a new getFirstAPlusCohortId() accessor.

SeapodymCohortDependencyAnalyzer.h/.cpp: rewrote the A+ branch of the constructor; updated doc comments/diagram; added getFirstAPlusCohortId().
tests/testTaskStepFarmingCohortAPlus.cxx: replaced sign-based checks with the new ID threshold.
No behavior change: verified the new dependency graph is structurally identical to the old one (isomorphic under the old↔new A+ ID mapping) across several (numAgeGroups, numTimeSteps, ageMature) cases including edge cases.