chore: enable metrics by default

[JoshuaMoelans]

Metrics captured by the Public API should be opt-out as per develop docs.

Since we don't have any internal auto-instrumented metrics, we enable it by default so calling sentry_metrics_X will just capture instead of requiring setting the option. ⚠️ this does make it so our batching thread starts regardless of whether it actually has any work (see the discussion below).

Also updates the tests to reflect this change; by default it is true, but one can set it to false with sentry_options_set_enable_metrics(options, false); (for example, when only wanting metrics for debug and not production).

follow-up

• update https://docs.sentry.io/platforms/native/metrics/
• the same for logs once that goes opt-out as well.

---

Discussion

CI currently fails because we now log from our Metrics Batcher Thread that starts by default; we can fix this either by

a) changing those logging tests to expect the output (but then starting the metrics batcher even if no metric is ever sent) #1612
b) adding a lazy init to the metrics/logs batcher #1611

• think about whether we also want the metrics batcher to always be running, since enable_metrics just starts it; maybe we only want to do so once the first metric is captured? ⚠️ currently, if we don't do this, this changes the behavior of the SDK (a batching thread starts even if metrics wasn't explicitly enabled, since it is now on by default)
  • see feat: add lazy init to batcher #1611

[JoshuaMoelans]

CI currently fails because we now log from our Metrics Batcher that starts by default; we can fix this either by

a) changing those logging tests to expect the output (but then starting the metrics batcher even if no metric is ever sent)
b) adding a lazy init to the metrics/logs batcher #1611