The delayed ftdc stat tells us when FTDC has a gap but it only ever assumes one collection stalled. Consider an example with FTDC running with a 1 second collection period where FTDC stalled for 1 second and another example where it stalled for 10 seconds. Ideally, the first case would report 1 delayed collection and the second case would report 10, but in reality we report 1 for each. This makes it confusing to interpret the stat since we don't get insight into how bad the stall really was. Furthermore, this misleading stat may lead us to make incorrect assumptions about a nodes health.

We should add a new stat here which considers the FTDC collection period when incrementing.