metrics: add process_uptime_seconds_total metric

[hawkw]

The proxy exposes a process_cpu_seconds_total metric. It also exposes
a process_start_time_seconds metric. But given a random metrics
snapshot, we have no way of knowing the proxy's uptime unless we know
the current time at the time of the snapshot. linkerd/linkerd2#8443
proposed adding a process_uptime_seconds_total metric tracking the
(fractional) number of seconds since the process started, to make these
other process metrics easier to interpret.

This branch implements the process_uptime_seconds_total metric.
Actually adding the metric is rather simple.

However, it occurred to me that uptimes are also used as timestamps in
the proxy's tracing logs. Currently, the start time used to calculate
log timestamps is just recorded when constructing the logger. If
separate Instants are used as the start time for log uptime timestamps
and for the process_uptime_seconds_total metric, this means the
process' uptime as reported by the metrics endpoint and the process'
uptime according to log messages may not quite agree. Therefore, I
thought it would be good if a single Instant was used for both logging
and metrics, and I made some additional changes so that we can record a
single timestamp when the proxy starts, and pass it into both the
metrics endpoint and the logger. This does make this branch a slightly
larger change, though.

Closes linkerd/linkerd2#8443

[olix0r]

IIUC this is only called in one place. Should it just take an Instant?

[hawkw]

good call, changed it to just pass in the Instant there!