Changing uptime metric_type from counter to gauge across all packages for consistent mapping

[ritalwar]

There was a big discussion about how to categorize the uptime metric in the Windows package i.e whether it should be classified as a "gauge" or a "counter."

Even though uptime usually acts like a counter by going up over time, the discussion considered "temporality." After talking it through, everyone agreed that no matter if we're using "delta" or "cumulative" temporality settings, we should always treat uptime consistently and decided that uptime should never be reported as a "delta." So, labeled uptime as a "gauge." in Windows package.

While uptime is reported as a gauge in many packages, there are a few where it's reported as a counter. Therefore, we should update the current integrations to maintain a consistent mapping. Here's a list of packages that currently use uptime as a counter and need updates:

• Apache
• Elastic Package Registry
• System (uptime datastream)
• AWS (RDS)

[constanca-m]

There is a related issue on delta metrics: elastic/elasticsearch#98171.