Periodic snapshot refresh

[dannimad]

This change triggers fetching the latest snapshot in the serializedStateManager after a timeout established by config provider. Once that is complete, ssm will try to update the base snapshot if the already given conditions are met. Such timer is initialized in three different scenarios:

1. While setting the initial snapshot while attaching the container.
2. While loading a container with no pending state.
3. Once we updated the snapshot for a subsequent refresh.

The scenario in which we load from a container with pending state is already covered by current code.

Here are some scenarios this change is covering, explaining its behavior:

1. Start fetching once timer is finished, fetching succeeds, conditions to refresh are met, base snapshot is updated, and a timer is restarted for the next attempt. On the next attempt the same procedure is followed.
2. Start fetching once timer is finished, fetching succeeds, conditions to refresh are not met, wait for all conditions to be met and attempt again via saved event, snapshot is updated.
3. Start fetching once timer is finished, fetching succeeds, conditions to refresh are met, snapshot is not updated due to being too old to refresh, timer restarts.
4. Start fetching once timer is finished, fetching does not succeed, attempt again until it succeeds.

Current conditions to update the snapshot should be sufficient since the most complicated case (while loading with pending state) is already covered.

[msfluid-bot]

Baseline CI build failed, cannot generate bundle analysis at this time

---

Baseline commit: d976469

Generated by 🚫 dangerJS against e6a9d5a

[anthony-murphy]

is there anything preventing this from overlapping with the timer? do we do anything make sure only 1 tryRefreshSnapshot is ever happening at once?

[anthony-murphy]

i would probably just skip this, and rely on the driver to retry and suceed, and if it fails it will happen again when the timer triggers again,

[dannimad]

I believe by removing this tryRefreshSnapshot and blocking when latestSnapshot is set should be enough to prevent overlapping but I can wrap this function on a runSingle

[dannimad]

I believe the runSingle approach doesn't work since tryRefreshSnapshot is not async. Conditions should be sufficient to prevent reentrancy.