feat: [NPM] metric for total Pod IPs

[huntergregory]

Reason for Change:
Windows has scale limits in terms of total Pod IPs in the cluster (on either Windows or Linux nodes).

Issue Fixed:

Requirements:

• uses conventional commit messages
• includes documentation
• adds unit tests

Notes:
Also removes unnecessary metric for max ipset members. Total Pod IPs is a more accurate/relevant metric.

Metric Preview

# HELP npm_pods_watched Number of Pods NPM tracks across the cluster including Linux and Windows nodes
# TYPE npm_pods_watched gauge
npm_pods_watched 22

[huntergregory]

Goal: call metrics.AddPod() once per Pod/IP pair (similar for metrics.RemovePod())

• If a Pod has an update to its IP, follow Pod DELETE then Pod CREATE.
• For Pod DELETE: c.cleanupDeletePod() is only called when podKey exists in c.podMap
• For Pod CREATE: c.syncAddedPod() is only called when podKey does not exist in c.podMap. Otherwise, another codepath is taken.
• Keys are only added/removed from c.podMap in c.syncAddedPod() and c.cleanupDeletePod()