Merge failed and successful job history limits for CronWorkflows into a single limit

# Summary

Argo Workflows, in its current state, allows the separate specification of the maximum number of failed and successful jobs to keep around for CronWorkflows (by setting `spec.failedJobsHistoryLimit` and `spec.successfulJobsHistoryLimit` respectively). Setting `spec.failedJobsHistoryLimit` to too low a number sometimes puts one in a position where a certain failed job is no longer available for inspection. On the other hand, setting it to too high a number results in a large number of failed jobs being left around, long after the workflow in question has moved to a state where it's consistently succeeding.

May I, therefore, request that `spec.failedJobsHistoryLimit` and `spec.failedJobsHistoryLimit` be merged into a single limit that keeps the last `k` failed _OR_ successful jobs?

I'm running v3.0.7 of Argo Workflows.

# Use Cases

Situations where one needs to debug a failing CronWorkflow.

---

**Message from the maintainers**:

Impacted by this bug? Give it a 👍. We prioritise the issues with the most 👍.

Start by locating the CronWorkflow definitions and the code that uses spec.failedJobsHistoryLimit and spec.successfulJobsHistoryLimit; the issue names no files or tests. Trace the history-pruning behavior and determine how one combined limit should retain the last k failed or successful jobs. Done means the unified behavior is implemented and covered by relevant tests.

Written by the indexing model from the issue text.