fix: honor updated TTL when overwriting feature cache entries

[fallintoplace]

What

• Honor the new TTL when overwriting a feature cache entry.

Why

• The old TTL was reused. Shorter TTLs kept stale features, and longer TTLs expired fresh features too soon.

Implementation

• Set the entry TTL before updating its value and expiry.
• Keep method signatures unchanged. Add regression cases for changed, zero and unchanged TTLs, expired entries, and separate keys.

[greptile-apps]

Confidence Score: 4/5

[Medium risk] Fixes feature cache TTL refresh on overwrite.

The PR should not merge until concurrent refreshes preserve the TTL belonging to each cached value.

_{Reviews (1) · Last reviewed commit: "fix: honor updated TTL when overwriting ..."}

[greptile-apps]

Concurrent refreshes mix TTLs

When two clients refresh the same cache key at the same time with different TTLs, one can change entry.ttl before the other calls update(). For example, a refresh requesting a five-second TTL can store its value with a 600-second expiry instead. Those features may remain cached long after the requested five seconds. Keep each refreshed value paired with its own TTL and expiry.

Knowledge Base Used: Client lifecycle and feature refresh

Prompt To Fix With AI

This is a comment left during a code review.
Path: growthbook/growthbook.py
Line: 120

Comment:
**Concurrent refreshes mix TTLs**

When two clients refresh the same cache key at the same time with different TTLs, one can change `entry.ttl` before the other calls `update()`. For example, a refresh requesting a five-second TTL can store its value with a 600-second expiry instead. Those features may remain cached long after the requested five seconds. Keep each refreshed value paired with its own TTL and expiry.

**Knowledge Base Used:** [Client lifecycle and feature refresh](https://app.greptile.com/growthbook/-/custom-context/knowledge-base/growthbook/growthbook-python/-/docs/client-lifecycle-and-feature-refresh.md)

---

For each issue above, determine whether it is valid and should be fixed. If so, fix it directly.

[vazarkevych]

Heads-up: this overlaps with #114, which adds a threading.Lock around
InMemoryFeatureCache.get/set/clear. That lock makes the two-step
entry.ttl = ttl; entry.update(value) atomic, so it resolves the P1 greptile
raised here — but only if both land, and whichever merges second will conflict
in set().

An alternative that fixes it independently of #114, and keeps the signature
backwards compatible:

def update(self, value: Dict[str, Any], ttl: Optional[int] = None) -> None:
    if ttl is not None:
        self.ttl = ttl
    self.value = value
    self.expires = time() + self.ttl

Then set() stays a single call and each refreshed value is paired with its own
TTL by construction.