ref(track_outcome): use FutureTrackingProducer for everything

[bmckerry]

getsentry/arroyo#547 introduced a producer abstraction in Arroyo which is essentially the same as TaskProducer but can be used anywhere, not just in tasks.

This PR replaces TaskProducer with FutureTrackingProducer in track_outcome, and cleans up all of the rollout logic now that this is fully rolled out everywhere.

[sentry]

Bug: Replacing SingletonProducer with FutureTrackingProducer may cause outcome loss during server shutdown, as the new producer might close before OutcomeAggregator flushes its buffer.
_{Severity: HIGH}

Suggested Fix

Verify that FutureTrackingProducer registers its atexit handler at construction time, ensuring it closes after OutcomeAggregator flushes during shutdown. Consider adding a test to confirm this shutdown ordering and prevent data loss.

Prompt for AI Agent

Review the code at the location below. A potential bug has been identified by an AI
agent. Verify if this is a real issue. If it is, propose a fix; if not, explain why it's
not valid.

Location: src/sentry/utils/outcomes.py#L176-L183

Potential issue: The replacement of `SingletonProducer` with `FutureTrackingProducer`
may reintroduce a shutdown-ordering bug. The previous implementation used an eager
`atexit` handler to ensure the producer remained active until after `OutcomeAggregator`
flushed its buffered outcomes during server shutdown. It is unclear if
`FutureTrackingProducer` provides the same guarantee. If the new producer's shutdown
handler runs before `OutcomeAggregator._atexit_flush()`, any outcomes flushed by the
aggregator during shutdown will be silently lost because the producer will already be
closed.

_{Did we get this right? 👍 / 👎 to inform future reviews.}

[bmckerry]

FutureTrackingProducer has effectively the same behaviour as SingletonProducer on shutdown